The evolution and structure/function of bHLH–PAS transcription factor family

Abstract

Proteins that contain basic helix-loop-helix (bHLH) and Per-Arnt-Sim motifs (PAS) function as transcription factors. bHLH–PAS proteins exhibit essential and diverse functions throughout the body, from cell specification and differentiation in embryonic development to the proper function of organs like the brain and liver in adulthood. bHLH–PAS proteins are divided into two classes, which form heterodimers to regulate transcription. Class I bHLH–PAS proteins are typically activated in response to specific stimuli, while class II proteins are expressed more ubiquitously. Here, we discuss the general structure and functions of bHLH–PAS proteins throughout the animal kingdom, including family members that do not fit neatly into the class I-class II organization. We review heterodimerization between class I and class II bHLH–PAS proteins, binding partner selectivity and functional redundancy. Finally, we discuss the evolution of bHLH–PAS proteins, and why a class I protein essential for cardiovascular development in vertebrates like chicken and fish is absent from mammals.

Introduction

The basic helix-loop-helix Per-Arnt-Sim family of proteins (bHLH–PAS) are transcription factors that regulate diverse biological functions (Figure 1). bHLH–PAS proteins are present throughout the animal kingdom and contain several conserved functional domains: a bHLH domain at the N-terminus that facilitates binding to DNA, two tandem PAS domains (PAS-A, PAS-B), and a domain at the C-terminus that activates or blocks transcription of target genes (Figure 1B). bHLH–PAS proteins are divided into two classes based on their dimerization properties. Class I bHLH–PAS proteins are typically specialized transcription factors that are activated in response to a stimulus such as a small molecule ligand, oxygen levels or neuronal activity. Class II bHLH–PAS proteins, which are often widely expressed, dimerize with class I proteins to form a functional transcriptional complex. The expression patterns of class I and class II proteins, together with the stimulus-dependent activation of class I proteins, allows bHLH–PAS protein heterodimers to activate or repress target genes in a spatially and temporally distinct manner. In this article, we will review the general structure and functions of the bHLH and PAS domains from bHLH–PAS proteins throughout the animal kingdom, focusing on animals commonly used in biomedical research. We highlight bHLH–PAS family members that contain only one PAS domain and do not fit neatly into the class I-class II organization. We discuss heterodimerization between class I and class II bHLH–PAS proteins, binding partner selectivity and functional redundancy. Some class II bHLH–PAS proteins may form homodimers, while some class I proteins may have functions other than as transcription factors. Finally, we discuss the evolution of bHLH–PAS proteins, and why a class I protein essential for cardiovascular development in vertebrates like chicken and fish is absent from mammals.

Figure 1.

Structure and function of bHLH– PAS proteins.(A) bHLH–PAS transcription factors act as heterodimers, where a class I and indicated class II protein interact and bind DNA. Asterisks indicate proteins encoded by genes present in vertebrates but absent from mammals. Proteins in the indeterminate class can bind class I proteins or act independently of class I and class II proteins. (B) Cartoon structure of representative bHLH–PAS proteins indicating key functional domains: basic helix-loop-helix (bHLH), Per-Arnt-Sim motifs (PAS-A, PAS-B), transcriptional activation or repression domains (TAD, TRD), LxxLL motifs, which modulate protein–protein interactions that influence transcription.

Functions of bHLH–PAS proteins

bHLH–PAS proteins function as transcription factors that regulate gene expression. bHLH–PAS proteins regulate gene expression in many cell types, during multiple stages of organismal development and adulthood (Table 1). Aryl hydrocarbon receptor (AHR) is a ligand-dependent transcription factor that is activated by environmental hydrocarbons and endogenous ligands such as tryptophan metabolites [1,2]. AHR up-regulates enzymes that metabolize and detoxify environmental small molecules. AHR also regulates the development and function of the liver, intestine and the immune system [3–8]. Hypoxia inducible factors (HIF1α, HIF2α, HIF3α) are activated by hypoxia and regulate the growth of blood vessels and differentiation of blood cells [9–17]. Circadian Locomotor Output Cycles Kaput (CLOCK), neuronal PAS domain protein 2 (NPAS2), aryl hydrocarbon receptor nuclear translocator like proteins (ARNTL, ARNTL2, also known as brain and muscle ARNT-like 1 and 2 or BMAL1, BMAL2) regulate circadian rhythms [18–26]. Neuronal PAS domain proteins 1, 3 and 4 (NPAS1, NPAS3, NPAS4) and SIM bHLH transcription factors 1 and 2 (SIM1, SIM2) regulate neuronal development and synaptic plasticity [27–34]. Aryl hydrocarbon receptor nuclear translocator proteins (ARNT, ARNT2) are essential coregulators for class I bHLH–PAS transcription factors [33,35–43] (discussed below), while nuclear receptor co-activators (NCOA1, NCOA2, NCOA3, also known as SRC-1, −2, −3) are essential coregulators for nuclear hormone receptors [44–52] (Figure 1).

Table 1.

List of bHLH-PAS genes and their functions from animals commonly studied for biomedical research

Species	bHLH–PAS gene name	bHLH–PAS protein name	Protein class	Function of protein	Reference (PMID)
Caenorhabditis elegans	aha-1	AHA-1	II	Mutation is larval lethal Neuronal specification with ahr-1 Hypoxic transcriptional response with hif-1 Ortholog of human ARNT/ARNT2	14757639 19632181
Mus musculus	Ahr	AHR	I	Angiogenesis and hematopoiesis Xenobiotic metabolism Development and function of the immune system Immune checkpoint regulation Regulation of viral replication Liver development and function	15590894 16443691 33941684 7732381 8806883 34446714 33408169 33239445 32690969 27783946 27158906 26005855 21068375 20676095 20676092
Homo sapiens	AHR	AHR	I	Angiogenesis and hematopoiesis Xenobiotic metabolism Immune response Mutation associated with retinitis pigmentosa	21976023 21999944 23723449 29726989
Caenorhabditis elegans	ahr-1	AHR-1	I	Neuron development Aggregation on lawns of bacterial food Lifespan	33349622 24655420 16919260 15136141 14757639
Danio rerio	ahr1a	Ahr1a	I	Unclear ( possible role in transcription in endothelial cells)	30907958 28817646
Danio rerio	ahr1b	Ahr1b	I	Unclear ( possible role in transcription in endothelial cells)	30907958 28817646
Danio rerio	ahr2	Ahr2	I	Xenobiotic metabolism Immune response Fin development Vascular morphology Craniofacial development	30907958 22242167 28817646 32384158 29494622
Mus musculus	Ahrr	AHRR	indeterminate	Regulate AHR activity Modulate xenobiotic metabolism	17949687 9887096
Homo sapiens	AHRR	AHRR	indeterminate	Represses AHR transcription activity Possible tumor supressor in cancers	17890447 18172554
Danio rerio	ahrra	Ahrra	indeterminate	Modulate Ahr2 activation by dioxins Possible role in hematopoiesis	19494032 24675095 18000031
Danio rerio	ahrrb	Ahrrb	indeterminate	Modulate Ahr2 activation by dioxins	19494032 18000031
Mus musculus	Arnt	ARNT (HIF1β)	II	Binding partner of multiple Class I bHLH–PAS transcription factors	27782878 26245371 28602820 7592839 17023418
Homo sapiens	ARNT	ARNT (HIF1β)	II	Binding partner of multiple Class I bHLH–PAS transcription factors	8662957 27782878
Danio rerio	arnt	Arnt (Hif1β)	II	Binding partner of multiple Class I bHLH–PAS transcription factors	16306231 16936225 Edwards et al. bioRxiv 2022
Mus musculus	Arnt2	ARNT2	II	Binding partner of multiple Class I bHLH–PAS transcription factors	8657146 10873592 31585079
Homo sapiens	ARNT2	ARNT2	II	Binding partner of multiple Class I bHLH–PAS transcription factors Mutation causes post-natal microcephaly, visual and renal anomalies	24022475 27782878
Danio rerio	arnt2	Arnt2	II	Binding partner of multiple Class I bHLH–PAS transcription factors	19374551 19234064 16936225 Edwards et al. bioRxiv 2022
Mus musculus	Arntl	ARNTL (BMAL1)	II	Binding partner of CLOCK and NPAS2 Circadian rhythm regulation	17417633 9160755 9616112
Homo sapiens	ARNTL	ARNTL (BMAL1)	II	Binding partner of CLOCK and NPAS2 Circadian rhythm regulation	17417633 9616112 27782878
Danio rerio	arntl1a	Arntl1a	II	Circadian rhythm regulation with clock genes Ortholog of human ARNTL	23468616
Danio rerio	arntl1b	Arntl1b	II	Circadian rhythm regulation with clock genes Ortholog of human ARNTL	23468616
Mus musculus	ARNTL2	ARNTL2 (BMAL2)	II	Binding partner of CLOCK and NPAS2 Circadian rhythm regulation	11207387 20153195
Homo sapiens	ARNTL2	ARNTL2 (BMAL2)	II	Binding partner of CLOCK and NPAS2 Circadian rhythm regulation	10964693
Danio rerio	arntl2	Arntl2 (Bmal2)	II	Circadian rhythm regulation	11517315
Caenorhabditis elegans	cky-1	CKY-1	I	Speculated to be functionally similar to NPAS4	14701734 19284974
Drosophila melanogaster	clk	Clock	I	Circadian regulation with cyc Regulation of circadian genes per and tim	11106063
Mus musculus	Clock	CLOCK	I	Circadian rhythm regulation Circadian regulation of genes in multiple tissues	17417633 9160755 9616112
Homo sapiens	CLOCK	CLOCK	I	Circadian rhythm regulation Circadian regulation of genes in multiple tissues	17417633 9616112
Danio rerio	clocka	Clocka	I	Circadian rhythm regulation Mesoderm patterning	18800057 28687631 23468616 18569451
Danio rerio	clockb	Clockb	I	Speculated function circadian rhythm regulation	23468616
Drosophila melanogaster	cyc	Cycle	II	Circadian regulation with clk Sensitization to cocaine	11106063 10409723 10446052
Drosophila melanogaster	dysf	Dysfusion	I	Required for tracheal migration, adhesion and fusion during development Joint formation in legs	16914738 12897136 17652079 25329825 30080872
Drosophila melanogaster	gce	Germ-cell expressed bHLH–PAS	indeterminate	Responds to juvenile hormone Can form heterodimer with met	26161662 16516852
Caenorhabditis elegans	hif-1	HIF-1	I	Stimulates gene expression in response to hypoxia Essential for heat acclimation	11427734 30696318 12686697
Mus musculus	Hif1a	HIF1A	I	Up-regulates genes in response to hypoxia Hematopoiesis	24141110 9436976
Homo sapiens	HIF1A	HIF1A	I	Regulates genes in response to hypoxia Vascular devleopment	9436976 9782081
Danio rerio	hif1aa	Hif1aa	I	Up-regulates genes in response to hypoxia Hematopoesis Neural protective	29339404 31314708
Danio rerio	hif1ab	Hif1ab	I	Up-regulates genes in response to hypoxia Hematopoesis Maturation of intestinal goblet cells	29339404 31314708
Danio rerio	hif1al	Hif1al	I	Up-regulates genes in response to hypoxia Regulates erythropoiesis Ortholog of human HIF3A	33037038 23034716 26052946
Danio rerio	hif1al2	Hif1al2	I	Maturation of intestinal goblet cells Speculated up-regulates genes in response to hypoxia	27222589
Mus musculus	Epas1	HIF2A	I	Regulates gene expression in response to hypoxia Hematopoesis Regulates vascular endothelial growth factor (VEGF) Spermatogenesis	15626745 20181618 15851592
Homo sapiens	EPAS1	HIF2A	I	Regulates gene expression in response to hypoxia Vascular development Mutation causes familial erythrocytosis	15851592 18184961
Danio rerio	epas1a	Epas1a	I	Up-regulates genes in response to hypoxia hematopoesis ortholog of human EPAS1 (HIF2A)	29339404
Danio rerio	epas1b	Epas1b	I	Up-regulates genes in response to hypoxia Hematopoesis ortholog of human EPAS1 (HIF2A)	29339404
Mus musculus	Hif3a	HIF3A	I	Inhibits/represses expression of hypoxia-induced genes	25626335 9840812 11734856
Homo sapiens	HIF3A	HIF3A	I	Modulates expression of hypoxia-induced genes Dominant negative regulator of HIF1A	11573933 16126907 19694616 9840812
Caenorhabditis elegans	hlh-34	HLH-34	I	Unknown, expressed in a subset of interneurons Ortholog of human NPAS3	34604715
Drosophila melanogaster	met	Methoprene-tolerant	indeterminate	Responds to juvenile hormone Mutants are resistant to insecticide methoprene Can form heterodimer with Germ-cell expressed	1592245 26161662 9501163 15720391 16516852
Mus musculus	Ncoa1	NCOA1	indeterminate	Transciptional coactivator of steroid hormone receptors Energy metabolism and obesity Possible regulator of hematopoiesis	8616895 7481822 12954634 12507421 27958775
Homo sapiens	NCOA1	NCOA1	indeterminate	Transcriptional coactivator of steroid hormone receptors Transcriptional coactivator of STAT5 transcription factor	9427757 7481822 9223431 12954634 9252329
Danio rerio	ncoa1	Ncoa1	indeterminate	Transcriptional coactivator of steroid hormone receptors Hematopoiesis	27958775 30448381 27156127
Mus musculus	Ncoa2	NCOA2	indeterminate	Transcriptional coactivator of steroid hormone receptors Energy metabolism and obesity Fertility	9238002 8670870 12507241 12138202
Homo sapiens	NCOA2	NCOA2	indeterminate	Transcriptional coactivator of steroid hormone receptors	9238002 8670870
Danio rerio	ncoa2	Ncoa2	indeterminate	Possible regulator of hematopoiesis Transcriptional coactivator of steroid hormone receptors	18295965 18248177
Mus musculus	Ncoa3	NCOA3	indeterminate	Transcriptional coactivator of steroid hormone receptors Important for hearing Regulates adipogenesis hematopoiesis Inflammation	33326993 9765300
Homo sapiens	NCOA3	NCOA3	indeterminate	Transcriptional coactivator of steroid hormone receptors Mutation associated with progressive hearing loss	33326993 9765300
Danio rerio	ncoa3	Ncoa3	indeterminate	Development and function of otoliths Proposed trancriptional coactivator	33326993
Mus musculus	Npas1	NPAS1	I	Neural development Behavior	15347806 9012850 28499489
Homo sapiens	NPAS1	NPAS1	I	Neuronal devleopment Regulates genes involved in psychiatric diseases	9012850 28499489
Danio rerio	npas1	Npas1	I	Expressed in adult telencephalon Speculated to be involved in neural development	25556858
Mus musculus	Npas2	NPAS2	I	Circadian rhythm regulation	11441147 26895328 11163178 17417633
Homo sapiens	NPAS2	NPAS2	I	Circadian rhythm regulation Mutation associated with nonobstructive azoospermia	11441147 11163178 25956372
Danio rerio	npas2	Npas2	I	Not characterized	none
Mus musculus	Npas3	NPAS3	I	Neural development behavior Lung development and function	15347806 28499489 19581591
Homo sapiens	NPAS3	NPAS3	I	Neural development Mutations associated with schizophrenia	12746393 21709683 22228753 15924306
Danio rerio	npas3	Npas3	I	Insufficent data Speculated to be involved in neuronal development	-
Mus musculus	Npas4	NPAS4	I	Synaptic plasticity Memory	24201284 22194569 18815592
Homo sapiens	NPAS4	NPAS4	I	Variants associated with intellectual disability	33758288 24465693
Danio rerio	npas4a	Npas4a	I	Forebrain development Angiogenesis of intersegmental vessicles	25538572 28082451
Danio rerio	npas4b	Npas4b	I	Angiogenesis of intersegmental vessicles	28082451
Danio rerio	npas4l	Npas4l	I	Master regulator of hemato-vascular specification Formation of lens cataract	27411634 7588049 30861003
Drosophila melanogaster	sim	Single-minded	I	Midline central nervous system development Axonal extension	3345560 3345559 2242162 1760843 16439478
Mus musculus	Sim1	SIM1	I	Neuronal development Hypothalamus development and function Obesity	9784500 16291793 12947113 10587584
Homo sapiens	SIM1	SIM1	I	Obesity Neuroendocrine development	10587584 16924270 33434169
Danio rerio	sim1a	Sim1a	I	Hypothalamus development and function Axonal guidance	19234064 23222439 16691572
Danio rerio	sim1b	Sim1b	I	Hypothalamus development and function	19234064 16691572
Mus musculus	Sim2	SIM1	I	Development and function of neuroendocrine cells Axon guidance Overexpression causes Down syndrome phenotypes Growth and integrity of skeleton	14988428 12024028 16291793 10400987 10915774
Homo sapiens	SIM2	SIM2	I	Promotes glioblastoma cell invasion Part of Down syndrome critical region on chromosome 21	7647800 7568099 20448453 15946822
Danio rerio	sim2	Sim2	I	Not characterized	none
Drosophila melanogaster	sima	Similar	I	Hypoxic response Trachea development Ortholog of human HIF1A	12215541 19285057
Drosophila melanogaster	ss	Spineless	I	Distal limb formation Mutants have reduced sensory bristles Ortholog of human AHR	10433921 17084833 9573046
Drosophila melanogaster	tgo	Tango	II	Neuronal devleopment Glial migration Trachea development Limb formation Coregulator of class I genes Ortholog of human ARNT/ARNT2	8557198 10433921 9409674
Drosophila melanogaster	trh	Trachealess	I	Development of salivary duct Development of trachea distal limb Development Most similar to human NPAS1, NPAS3	17187773 8557198

Some bHLH–PAS proteins have additional functions beyond transcription. AHR is not only a transcription factor, AHR also functions as an E3 ubiquitin ligase that targets proteins for degradation [53]. NCOA1 is a histone acetyl transferase and increases the accessibility of chromatin for transcription [54]. It remains to be seen whether other bHLH–PAS proteins have additional functions beyond transcription.

Functional motifs in bHLH–PAS proteins

There are hundreds of proteins that contain either a bHLH domain or PAS domains [55–57]. bHLH domains bind DNA [58,59]. PAS domains detect a plethora of stimuli and regulate diverse biological processes throughout the animal kingdom, like nitrogen fixation in bacteria [60], growth in response to light in plants [61], circadian rhythms in fungi, insects, and vertebrates [18,62–65], and ion channel gating in vertebrates [66].

We define bHLH–PAS proteins as any protein containing both a bHLH domain and a PAS domain, of which there are 19 in humans. Most bHLH–PAS proteins contain a bHLH domain, two PAS domains, and a transrepression or transactivation domain (Figure 1B). Some bHLH–PAS proteins contain one PAS domain instead of two (Figure 1). bHLH–PAS proteins are divided into two classes, where class I proteins form dimers with class II proteins. There are only four class II proteins: ARNT (also known as HIF1β), ARNT2, ARNTL (BMAL1), ARNTL2 (BMAL2) versus a dozen class I proteins. The same class II protein can dimerize with multiple different class I proteins. Dimerization is restricted: class II proteins ARNTL and ARNTL2 appear to dimerize only with class I proteins CLOCK and NPAS2, while class II proteins ARNT and ARNT2 appear to dimerize with all other class I proteins except for CLOCK and NPAS2 [37] (Figure 1).

Some bHLH–PAS domain proteins contain only a single PAS domain, such as the aryl hydrocarbon receptor repressor (AHRR) and NCOA1, NCOA2 and NCOA3 (Figure 1). These proteins do not fit neatly into the class I and II dichotomy. AHRR behaves like a class I protein, in that it can bind ARNT and DNA and repress transcription [67]. However, AHRR proteins with mutations that prevent DNA binding or prevent dimerization with ARNT still retain the ability to repress AHR activation [68], suggesting that AHRR may block AHR activity using additional mechanisms besides forming a complex with ARNT and DNA.

It is also difficult to classify NCOA proteins as class I or class II. NCOA1, NCOA2 and NCOA3 proteins were originally identified as coregulators of nuclear hormone receptors, ligand-dependent transcription factors that are not members of the bHLH–PAS family [44,48,49,51]. NCOA1–3 act as coregulators for a variety of transcription factors, including class I bHLH–PAS proteins like the aryl hydrocarbon receptor [69–71]. Whether NCOA1–3 proteins form a transcriptional complex in the absence of class II bHLH–PAS proteins is not clear. NCOA1 and NCOA2 are associated with ARNT at the promoter region of CYP1A1, an AHR target gene, while NCOA1 was required for AHR-dependent reporter gene expression [70]. Overexpression of NCOA2 reduced expression of HIF1α-dependent reporter gene by competing with HIF1 α to bind ARNT [72]. These studies suggest that NCOA proteins bind ARNT, typical of class I proteins. However, in many cases where NCOA proteins form a transcriptional complex with nuclear receptors, there is absence of evidence as to whether ARNT or other class II bHLH–PAS proteins are part of the transcriptional complex. Mice and worms with mutations in Arnt (or aha-1, the C elegans ortholog) do not survive beyond embryo or larval stages [73–75], so it is difficult to test whether nuclear steroid receptor function, which requires NCOAs, is abnormal in the absence of ARNT proteins.

Class I and Class II dimerization selectivity

For a given class I protein, what regulates its dimerization with ARNT vs ARNT2 (or ARNTL vs ARNTL2)? What are the functional differences, if any, when a class I protein dimerizes with ARNT vs ARNT2? Alternatively, do ARNT and ARNT2 act redundantly? Below we provide examples using different class I proteins. Redundant functions of ARNT and ARNT2 may depend on several factors that remain to be explored, such as cell type, developmental stage, and identity of the Class I protein.

NPAS4 forms stimulus-dependent dimers with ARNT or ARNT2 in the brain

NPAS4 regulates the structure and strength of synapses in the nervous system and is required for contextual memory formation in mice [29,76,77]. NPAS4 is expressed following two different stimuli, action potentials and excitatory post synaptic potentials. In the hippocampus, when NPAS4 was induced by action potentials, it formed a complex with ARNT2 but not with ARNT or BMAL1, two other class II proteins expressed in the hippocampus. In contrast, when NPAS4 was induced by excitatory post synaptic potentials, NPAS4 formed a complex with ARNT exclusively [43]. NPAS4–ARNT and NPAS4–ARNT2 heterodimers bind different regions of the genome, suggesting that they target expression of different genes. NPAS4–ARNT was enriched at enhancer regions, while NPAS4–ARNT2 and NPAS4–ARNT were present at gene promoters [43].

Arnt and Arnt2 act redundantly with HIF1α and NPAS4L

In cultured mouse and human cell lines (Hepa1c4, HeLa), ARNT and ARNT2 appeared redundant in their ability to drive the expression of a hypoxia response element (HIF-dependent) reporter gene [78]. In zebrafish embryos, Npas4l is required for the specification of blood cells and most blood vessels (endothelial cells). Zebrafish embryos with mutations in both arnt and arnt2 genes, but not single arnt or arnt2 mutants, exhibit the same phenotype as npas4l mutants, suggesting that ARNT and ARNT2 act redundantly with NPAS4L [79].

AHR preferentially dimerizes with ARNT

The aryl hydrocarbon receptor (AHR) is a ligand-dependent class I protein. In zebrafish embryos, exposure to the AHR ligand 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) causes cardiac toxicity. Arnt2 mutants are sensitive to TCDD-dependent cardiac toxicity, whereas arnt1 mutants have normal cardiac morphology following TCDD exposure [79]. These results demonstrate that in zebrafish, the aryl hydrocarbon receptor preferentially interacts with Arnt and not Arnt2 in response to TCDD. Similar results were obtained in cultured cells, where activation of an AHR-dependent reporter gene in response to the AHR ligand 3MC was robust in the presence of ARNT and low in the presence of ARNT2. Additionally, 3MC-dependent expression of the endogenous AHR target gene CYP1A1 was robust in the presence of ARNT, but absent from cells expressing ARNT2 [78]. Mutating a single amino acid in the PAS-B region of ARNT prevented 3MC-dependent reporter gene expression. This mutation had no effect on HIF1α reporter gene expression [78]. Together, these results suggest that AHR preferentially dimerizes with ARNT and that differences in the ARNT and ARNT2 PAS-B regions are responsible for the preferential dimerization of AHR with ARNT vs ARNT2.

It is likely that the degree to which class II proteins act redundantly, or form selective dimers with class I proteins, depends on cell type, developmental stage, and relative expression levels. Much remains to be learned about class I-class II dimerization selectivity.

Evidence that ARNT can function without other bHLH–PAS proteins

ARNT can form homodimers in vitro. ARNT homodimers bound E-box core DNA sequence in vitro, consistent with DNA binding properties of the ARNT bHLH domain, and drove reporter gene expression from an E-box-containing promoter, the adenovirus major late promoter [38,80]. Structural studies of the human ARNT PAS-B domain demonstrated that it can form a homodimer [81]. To what extent ARNT forms homodimers in vivo is not known. It is not known whether ARNT2 can form homodimers.

ARNT and ARNT2 may also act as transcriptional coregulators with non- bHLH–PAS transcription factors. ARNT and ARNT2 enhanced the transcriptional activity of nuclear estrogen receptors alpha and beta (ERα, ERβ) [82]. Formation of an ER–ARNT complex required the C-terminal domain of ARNT and did not require the bHLH or PAS domains [82]. This contrasts with the AHR–ARNT complex, which requires ARNT bHLH and PAS domains for dimerization and formation of a transcriptional complex. Moreover, ARNTL (BMAL1) was unable to act as an ER coregulator [82]. Reducing the amount of available ARNT, either by small interfering RNAs targeting ARNT or by activating AHR or HIF1α signaling to compete for binding to ARNT, reduced the transcription activity of ERs [83]. Together, these results suggest that ARNT acts as a transcriptional coregulator for nuclear estrogen receptors, separate from its role as a bHLH–PAS dimer. A caveat is that NCOA proteins might be present in ER–ARNT transcriptional complexes.

ARNT can function as part of the NF-kappaB protein complex to regulate gene expression. In cultured lymphoma cells, ARNT binds the NF-kappaB subunit protein RelB. ARNT binding to RelB enhanced the binding of the ARNT-RelB-p50 protein complex to DNA, which inhibited expression of target genes [84]. These results suggests that ARNT acts as a transcriptional coregulator for NF-kappaB. However, it is not known whether any class I bHLH–PAS proteins are present in the NF-kappaB complex with ARNT in lymphoma cells. Class I proteins such as AHR and HIF2α interact with RelA, RelB and other NF-kappaB proteins in other contexts [85–90], though it remains to be seen whether ARNT or other class II proteins are necessary for NF-kappaB function in these contexts.

Evidence that AHR can function without class II bHLH–PAS proteins

ARNT is considered an obligate binding partner of AHR [39]. However, recent studies suggest that under certain conditions AHR can act independently of ARNT to up-regulate gene expression. In mouse liver, TCDD-AHR was bound to the promoter of, and up-regulated, the Serpine1 gene, but no ARNT was detected at the promoter and knockdown of ARNT did not reduce expression [91]. In mouse hepatoma 1c1c7 cells, exposure to N-acetylsphingosine caused apoptosis in an AHR-dependent manner but reducing ARNT levels had no effect [92]. In a rat liver epithelial cell line (WB-F344), TCDD exposure increased JunD and cyclin A expression in an AHR-dependent manner, while knockdown of ARNT had no effect [93]. In human macrophages, AHR interacts with RelB to regulate the expression of several chemokine genes. Gel shift assays identified the promoter DNA sequence to which AHR and RelB bind, however ARNT was not detected in this complex [88]. Taken together, these results suggest that TCDD-AHR can up-regulate gene expression independently of ARNT. A limitation of these studies is that none considered ARNT2. ARNT2 was shown to interact with AHR and DNA in vitro [35,94,95], but to our knowledge there is an absence of evidence of whether ARNT2 interacts with AHR in vivo. DNA binding, coregulator recruitment and target gene expression could be different between AHR–ARNT and AHR–ARNT2 transcriptional complexes.

When AHR acts as a ubiquitin ligase, is ARNT required? In MCF7 breast cancer cells, knockdown of ARNT reduced AHR-dependent transcription but did not reduce AHR E3 ubiquitin ligase activity [96]. In contrast, in cultured HeLa cells overexpressing AHR, both AHR and ARNT were associated in a complex of ubiquitin ligase core proteins [53]. This result could be an artifact of AHR overexpression and not reflect what occurs in vivo. Alternatively, ARNT may act as a coregulator of AHR ubiquitin ligase activity in some cell types and not others.

Alternative splicing alters function of bHLH–PAS transcription factors

DNA is transcribed to produce precursor messenger RNA ( pre-mRNA). Portions of pre-mRNA are then excised or spliced out, leaving a mature messenger RNA (mRNA) that serves as a template for protein translation. The splicing process may occur such that the same pre-RNA produces multiple different mRNAs. Such alternatively spliced mRNAs may be translated into proteins with different functions, despite being transcribed from the same gene [97–100]. Alternative splicing occurs throughout the genome in many different cell types at all stages of development and adulthood. Aberrant splicing can cause or exacerbate diseases in humans [101–104]. How does alternative splicing influence the function of bHLH–PAS proteins? We know the most about alternative splicing of ARNT pre-mRNA.

Alternative splicing produces different ARNT proteins

The ARNT pre-mRNA is alternatively spliced to produce two different protein isoforms: ARNT isoform 1, which includes 15 amino acids encoded by exon 5, and ARNT isoform 3, which lacks exon 5 and those 15 amino acids [105]. The additional amino acids in isoform 1 include an amino acid phosphorylated by casein kinase 2, a phosphorylation site that is absent from isoform 3 [106].

In human multiple myeloma and anaplastic large cell lymphoma cell lines (KMS-18, Karpas 299), knockdown of the ARNT isoform 1 induced cell-cycle arrest and apoptosis [107]. Moreover, six different B- and T-cell cancer cell lines exhibited increased isoform 1 expression, or increased ratio of expression of isoform 1 to isoform 3, compared with normal human primary B and T cells [107]. Thus, isoform 1 confers a proliferation advantage.

How do cells regulate production of ARNT isoform 1 vs isoform 3? The RNA-binding protein RBFOX2 promotes the inclusion of exon 5 in the conversion of ARNT pre-mRNA to mRNA, leading to production of ARNT isoform 1 [108]. RBFOX2 gene expression was up-regulated in two human malignant T cell lines ( Jurkat, Karpas 299) compared with primary T cells. Knockdown of RBFOX2 expression led to a reduction in ARNT isoform 1 protein levels and a reduction in cell proliferation [108].

Do different ARNT isoforms influence the dimerization or function of class I proteins differently? This is best studied in the context of AHR, a ligand-dependent class I protein. In Karpas299 cells exposed to the AHR ligand TCDD, AHR target gene expression was different depending on the relative levels of ARNT isoform 1 vs isoform 3 [109]. Additionally, phosphorylation of the casein kinase 2 phosphorylation site, present exclusively in isoform 1, is required for optimal AHR activity [109].

These studies demonstrate clear functional differences between ARNT isoform 1 and ARNT isoform 3 proteins, differences that affect the function of one class I protein (AHR), cause differential target gene expression, and promote or reduce cell proliferation. Many exciting questions remain. How does alternative splicing of ARNT affect the function of other class I proteins besides AHR? Does alternative splicing of other class II proteins occur, and if so, how does it affect their function? To what degree is bHLH–PAS protein alternative splicing conserved in different species? ARNT isoforms 1 and 3 appear to be present in mice, although mice may have additional ARNT isoforms whose function is not known [110]. NPAS3 alternatively spliced forms appear to be similar in humans and chicken, however the functional differences, if any, are not known [111]. In zebrafish, there are three different isoforms of ARNT2. Overexpression of zebrafish AHR together with different ARNT2 isoforms in COS-7 monkey cells demonstrated that, in response to TCDD, one ARNT2 isoform up-regulated a TCDD reporter gene while the other two isoforms did not [94]. Meanwhile, all three isoforms induced gene expression similarly with a different class I protein, HIF2α [94], suggesting that ARNT2 isoforms affect the function of some, but not all, class I proteins. While it is not clear whether ARNT2 is similarly spliced in humans, the idea that alternative splicing affects bHLH–PAS protein function is supported by studies of ARNT isoforms in humans and ARNT2 isoforms in zebrafish.

Evolution of bHLH–PAS proteins

bHLH–PAS transcription factors containing PAS-A and PAS-B domains (eg canonical class I and class II proteins) are only found in metazoans [112]. Most invertebrate genomes contain less than a dozen bHLH–PAS genes, while vertebrate genomes contain dozens [112]. One gene is particularly interesting: npas4l, present in the genome of non-mammalian vertebrates such as fish but absent from mammals (Figure 2). In chicken and zebrafish embryos, Npas4l is essential for the development of blood cells and blood vessels [113–115]. How is such an essential gene absent from mammalian genomes? The closest related gene, Npas4, does not seem to be required for blood cell or blood vessel formation in mice. The functional homolog(s) of Npas4l in mammalian embryonic development are not known. One possibility is that, in the relatively hypoxic environment of the placenta, the expression of HIF genes changed in response to the low oxygen environment, and these class I bHLH–PAS proteins took over the role of Npas4l in specifying hemato-vascular progenitor cells. Zebrafish embryos develop outside of the mother, with free access to oxygen from the environment, via diffusion through embryonic tissue. In this high-oxygen environment, Npas4l is sufficient for driving hemato-vascular specification. In both zebrafish and mice, Arnt and Arnt2 genes are required for the proper development of blood cells and endothelial cells [74,79,116]. The same class II proteins might be used in all species, but the class I partners may have changed depending on evolutionary pressure.

Figure 2. Phylogenetic tree of bHLH–PAS proteins.

Unrooted phylogenetic tree of bHLH–PAS proteins from humans and commonly studied animal models of human development. The neighbor-joining tree was built using multiple alignments of the indicated bHLH–PAS amino acid sequences. To assess the degree of confidence of the nodes, bootstrap with 1000 repetitions was performed.

Perspectives.

• bHLH–PAS transcription factors regulate diverse biological processes throughout the body, from cell specification and differentiation in embryos to organ function in adults. Understanding the molecular mechanisms by which bHLH–PAS proteins regulate transcription is a prerequisite for understanding how bHLH–PAS function is abnormal in birth defects and disease.

• Current thinking indicates that bHLH–PAS proteins form heterodimers, where class I proteins interact with class II proteins. The class II proteins ARNT and ARNT2 interact with a variety of class I proteins, whereas the class II proteins ARNTL and ARNTL2 interact with only two class I proteins, CLOCK and NPAS2.

• Important future questions include to what degree class II proteins act redundantly and how specific bHLH–PAS proteins, such as NPAS4L, evolved to have essential roles in some animals but not in others.

Abbreviations

AHR

aryl hydrocarbon receptor

AHRR

aryl hydrocarbon receptor repressor

bHLH–PAS

basic helix-loop-helix Per-Arnt-Sim family of proteins

NPAS2

neuronal PAS domain protein 2

PAS

Per-Arnt-Sim motifs

TCDD

2,3,7,8-tetrachlorodibenzo-p-dioxin