Evolution of the enhancer-rich regulatory region of the gene for the cell-type specific transcription factor POU1F1

Abstract

Precise spatio-temporal expression of genes in organogenesis is regulated by the coordinated interplay of DNA elements such as promoter and enhancers present in the regulatory region of a given locus. POU1F1 transcription factor plays a crucial role in the development of somatotrophs, lactotrophs and thyrotrophs in the anterior pituitary gland, and in maintaining high expression of growth hormone, prolactin and TSH. In mouse, expression of POU1F1 is controlled by a region fenced by two CTCF sites, containing 5 upstream enhancer elements, designated E-A (5′ to 3′). Elements C, B and A correspond to elements shown previously to play a role in pituitary development and hormonal expression; functional roles for elements E and D have not been reported. We performed comparative sequence analysis of this regulatory region and discovered that three elements, B, C and E, are present in all vertebrate groups except Agnatha. One very long (>2 kb) element (A) is unique to mammals suggesting a specific change in regulation of the gene in this group. Using DNA accessibility assay (ATAC-seq) we showed that conserved elements in anterior pituitary of four non-mammals are open, suggesting functionality as regulatory elements. We showed that, in many non-mammalian vertebrates, an additional upstream exon closely follows element E, leading to alternatively spliced transcripts. Here, element E functions as an alternative promoter, but in mammals this feature is lost, suggesting conversion of alternative promoter to enhancer. Our work shows that regulation of POU1F1 changed markedly during the course of vertebrate evolution, use of a low number of enhancer elements combined with alternative promoters in non-mammalian vertebrates being replaced by use of a unique combination of regulatory units in mammals. Most importantly, our work suggests that evolutionary conversion of alternate promoter to enhancer could be one of the evolutionary mechanisms of enhancer birth.

1. Introduction

The key DNA sequence involved in regulation of gene expression is the promoter, located immediately upstream of the transcription start site. To this region bind RNA polymerase and transcription factors (TFs) involved in regulating gene activation and transcription. Additional regulatory DNA elements such as enhancers, are involved in precise tissue-, cell type- and time-specific expression of the gene in eukaryotes. Enhancers are short DNA sequences located upstream or downstream of their associated gene promoter; they bind TFs which interact in turn, in cis, with proteins bound to the promoter to execute their regulatory function. Enhancers are often located at a considerable distance from their target genes (many kb); their interaction with the promoter requires looping out of intervening DNA [1,2]. In the case of some genes, several well-spaced enhancer elements appear to function in a coordinated manner, and together are referred to as a super-enhancer; these are often associated with cell-type specific, highly-expressed genes, usually transcription factors that maintain cell identity [3]. Regulation of gene expression may also be achieved by use of more than one promoter, associated with different starting exons alternatively spliced to the rest of the gene [4].

Although conferring considerably different functions, promoters and enhancers share many structural and functional traits [5]. Both regions, when activated, display DNA accessibility, allowing different factors and co-factors to bind to their cognate sites. Both regions share several histone modifications (e.g. H3K4me1, H3K4me2, H3K27Ac) although at different levels. Moreover, both regions frequently bind the same factors, to exert their respective functions and both bind RNApolII which transcribes RNAs in both directions, although with significantly different efficiencies. While at the enhancer the levels of transcripts are low and rather equal in both directions, at the promoter the coding transcript is transcribed with much higher efficiency than the transcript going in the reverse direction. Taken together, enhancers and promoters share many common architectural and functional features, which may provide clues about the evolutionary origin of enhancers.

Here we examine the role played by promoter and enhancers associated with the POU1F1 gene, indispensable for the development of the anterior pituitary gland and for production of some of its hormones. Understanding of the regulatory role of these elements is incomplete and based largely on studies in rodents. Therefore, the present study was carried out to extend such understanding across all vertebrate groups.

POU1F1 (also known as Pit1) is a POU-family TF which plays a key role in regulating development of the anterior pituitary gland, and the expression of growth hormone (GH), prolactin and thyrotropin (TSH) [6,7]. It is expressed at high level in parts of the developing pituitary and in mature cells producing these three hormones (respectively somatotrophs, lactotrophs and thyrotrophs), but not at all in most other tissues, though low levels of expression in a few cell types may be of significance. Mutations leading to loss of POU1F1 function give rise to defective production of pituitary hormones and cause dwarfism in mice and humans [8], therefore a fuller understanding of its role in human pituitary disease is needed. The promoter immediately upstream of POU1F1 has been well-characterised in some mammals and clearly plays a crucial role in regulation of its expression [9]. In addition, an element 10 kb upstream of the promoter has been shown to be indispensable in maintaining high expression of POU1F1 in adult mouse tissue. Detailed studies using transgenic animals have shown the importance of several upstream elements in spatio-temporal expression of POU1F1 during pituitary development [[10], [11], [12]] including the long (>2 kb) element located ∼6 kb upstream of transcription start site (TSS), indispensable for early activation of the POU1F1 gene expression [13]. This element, called EEα, is bound by ATBF1, massive, multiple zinc-finger/homeodomain transcription factor, expressed specifically during early brain development. The identification of several discrete elements regulating the POU1F1 gene within 25 kb from the TSS raises the possibility that these comprise a super-enhancer.

Multiple enhancers and/or alternative promoters allow for complex and subtle regulation of gene expression. In the case of POU1F1 this includes precise regulation of the spatial and temporal expression of the gene during development, and the sustained high level of hormone expression maintained in mature pituitary somatotrophs, lactotrophs and thyrotrophs. The regulation of POU1F1 expression in rodents clearly involves multiple enhancers, but the extent to which this applies to other species is not clear. The objective of this study was to test the hypothesis that POU1F1 expression is governed by an enhancer-rich region in all vertebrates, by performing a comparative study exploring the sequences of the upstream region of this gene in a wide range of species, and DNA accessibility in some cases. The results are presented as the first broad comparative analysis of conserved elements upstream of the POU1F1 gene. The analysis involved representatives of all tetrapod groups, followed by a detailed examination of the alternative splicing and use of an alternative promoter revealed in birds and some amphibians and fish, detailed analysis of the enhancer-rich region in fish and mammals, analysis of CTCF elements and promoters and results of DNA accessibility studies in several non-mammalian vertebrates. The evolutionary implications of the results are discussed in detail, and phylogenetic trees are presented to summarize the conclusions drawn about evolution of POU1F1 regulation. We show how a conserved pattern of enhancers is used to regulate POU1F1 expression with specific modifications found in different clades. Our data show how in birds and some other non-mammalian vertebrates alternative splicing of upstream exons appears to play an important role in regulating the POU1F1 gene, but in placental mammals a more-complex use of enhancers applies and an alternative first exon is not employed.

2. Materials and methods

2.1. Sources of pituitaries

Tissues from guinea fowl (Numida meleagris), alligator (Alligator mississipiensis) and bullfrog (Lithobates catesbeianus) were a kind gift from Drs Daley and Azizi laboratories at UCI. Mouse colony (C57Bl6) is established in D. S–K laboratory. All procedures are performed under approved protocols.

2.2. Genomic sequences

Genomic sequences for POU1F1 were extracted from the ensembl or NCBI wgs databases, following BLAST searching [14] to locate the genes. In some cases additional searches of NCBI databases were used to extend/complete genomic sequences. Details for species included in this paper are given in Supplementary Table S1.

2.3. Transcriptomes

Pituitary transcriptomes for various mammals, birds, amphibia and fish were accessed through the NCBI SRA database, and used to assess transcription levels of POU1F1, and details of splicing patterns. Expression profiles were established as described previously [15].

2.4. ATAC-seq

ATAC-Seq was performed as previously published [16]. In brief, fresh tissue was re-suspended immediately in 1 ml ice cold nuclei permeabilization buffer (5%BSA, 0.2% (m/v) NP40, 1 mM DTT in PBS solution) with 1X complete EDTA-free protease inhibitor and homogenized using a syringe and needle followed by slow rotation for 10 min at 4 °C. The nuclei suspension was then filtered through a 40 μM cell strainer and centrifuged for 5 min at 500×g at 4 °C. The nuclei pellet was resuspended in an ice cold 50 μl tagmentation buffer. Nuclei concentration was adjusted to 2000–5000 nuclei/μl and 10 μl of the suspension was used for tagmentation. 0.5 μl Tagment DNA Enzyme 1 (FC-121-1030, Illumina) was added to the 10 μl suspension. The reaction mix was thoroughly pipetted and incubated 30 min with 500 rpm at 37 °C. After the tagmentation reaction completed, the DNA was isolated using Qiagen PCR Purification Kit (Cat.#.28304, Qiagen) and eluted in 20 μl Elution Buffer. The eluted DNA fragments were then amplified by PCR with Nextera compatible indexed sequencing i5 and i7 adapters using NEBNext 2x PCR Master Mix PCR kit (M0541, NEB). The amplified DNA library was fragment size selected from 200bp to 800bp using Ampure XP beads (A63880, Beckman Coulter). The quality of the ATAC-Seq libraries was assessed by Agilent 2100 bioanalyzer (Agilent Technologies, Inc.). ATAC-Seq libraries were pooled and run on a NovaSeq 6000 System (Flow Cell Type S4) Illumina sequencer with a paired-end read of 100 bp to harvest about 50 million paired-end reads per sample.

2.5. Identification of enhancer elements and POU1F1 binding sites

In many cases conserved enhancer elements were identified upstream of the POU1F1 gene in various species by BLAST searching of wgs databases with sequences from related organisms (see example in Supplementary Fig. S6). In other cases more-focussed searches of sequences upstream of the POU1F1 gene were necessary, using BLAST or clustalw [17]. Potential POU1F1 binding sites were identified using PROMO [18], with taxon restriction to Craniata.

2.6. Sequence alignments

Alignment of enhancer sequences was carried out using M-Coffee [19,20] followed by manual adjustment and evaluation using the TCS procedure [21].

2.7. Phylogenetic trees

The phylogenetic trees in Fig. 6 were constructed on the basis of phylogenies published previously [22,23].

Fig. 6.

Phylogenetic trees summarizing the main events in POU1F1 regulatory region evolution in (A) Vertebrates and (B) Mammals. Phylogenies are based on [22,23]. Presence or absence of regulatory elements A-E in selected species is indicated.

3. Results

3.1. The POU1F1 regulatory region in tetrapods

To visualize open regulatory elements in the Pou1f1 gene in mouse pituitary we have performed ATAC-seq on nuclei isolated from adult (5 month-old) male mice. DNA accessibility studies revealed a cluster of 5 open regions upstream of the mouse POU1F1 gene promoter, corresponding to five potential enhancers (elements E-A) (Fig. 1). Enhancer elements A, B and C have been characterized in detail previously and correspond to elements EEα, EEβ, and DE [[9], [10], [11], [12], [13],24]. Elements D and E are upstream of these. Highly conserved CTCF elements upstream of element E and in intron 5 flank the set of enhancers and coding region of the gene.

Fig. 1.

Organization of the POU1F1 gene in representative tetrapods. Black and grey bars represent the 6 (mammals) or 7 (other vertebrates) exons. Red bars indicate conserved sequences (A–E) corresponding to the accessible regions in the ATAC profile (top panel). Blue bars represent alternatively spliced exon(s) found in birds (see text and Fig. 2); in mouse and human alternative upstream exon(s) are not used, but in the other species shown their usage or not is unknown due to lack of pituitary transcriptomes. F indicates a sequence upstream of the CTCF motif that is conserved in most placental mammals but not other tetrapods; it is not associated with an ATAC peak, and is probably not involved in POU1F1 regulation. Sequences included are: MusMus, mouse, HomSap, human, MonDom, oppossum (mammals); GalGal, chicken, AnsCyg, swan goose, CorCor, crow (birds); AnoCar, anole lizard, AllMis, alligator (reptiles); LitCat, bullfrog, XenTro, clawed toad, RhiBiv, 2-lined caecilian (amphibia); LatCha, coelacanth (sarcopterygian fish). For full species names see Supplementary Table S1. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

To assess how widely the type of complex regulatory region seen in mouse occurs, POU1F1 genes in a wide range of tetrapod species were investigated, identifying conserved sequences by BLAST analysis. The five-enhancer arrangement was conserved in many mammals, including marsupials (Fig. 1), but in a number of mammalian groups one, or sometimes two enhancers are lost. In reptiles, birds and amphibia, equivalents to only three of the five enhancers seen in mouse are conserved. Elements A and D are absent in all cases examined, but Elements B, C and E are present in almost all cases, suggesting a less flexible arrangement than seen in mammals (Fig. 1). Some variation is apparent however. In particular, in Anura (frogs and toads) the orientation of element C is reversed compared with that in other amphibians and in all other tetrapods. Furthermore, the size of the regulatory region, and spacing of elements within it, is very variable. Most notably, in the axolotl, a salamander, the regulatory region extends over ∼221 kb, much greater than non-salamanders (cf 9–23 kb for the amphibia included in Fig. 1) and indeed other tetrapods examined. The introns of the axolotl POU1F1 gene are also exceptionally large, though the spacing between promoter and exon 1 is similar to that in other amphibia. The large gene reflects the very large axolotl genome [25]. Notably, the sequences of the enhancer elements and of coding sequence in axolotl are conserved and unremarkable (Supplementary Fig. S1) suggesting that the greatly increased gene size is not associated with functional changes. Finally, in human, enhancer element D is absent. A fuller assessment of the regulatory region across mammals is considered below.

3.2. Alternative splicing and alternative promoters

A clear-cut transition in POU1F1 regulation appears to have occurred during the course of evolution between mammals and birds, and potentially other non-mammalian tetrapods. In mammals transcription of POU1F1 is controlled by up to 5 enhancers, comprising a potential super-enhancer; there is no evidence for use of alternative promoters. Single cell studies [26] have shown that all 5 regulatory elements are accessible in each cell-type expressing the targets of POU1F1, further suggestive of cooperative work of all enhancers as a cluster in adult pituitary (Supplementary Fig. S2).

In non-mammalian tetrapods conserved sequences corresponding to just 3 of these enhancers are present. For a number of bird species pituitary transcriptomes are available. Analysis of these showed that in each case two alternative promoters are used, one equivalent to that seen in mammals (upstream of exon 1), the other equivalent to mammalian enhancer element E, upstream of an alternative exon, exon 0 (Fig. 2A and B). These two promoters are used in approximately equal proportions. Exon 0 and exon 1 are spliced alternatively to exon 2. Analysis of a number of available mammalian pituitary transcriptomes (eutherians, including mouse, human and pig) confirmed that there is no alternative upstream exon - almost all transcripts start with exon 1. Reptiles and amphibia resemble birds in having just the three conserved upstream elements; whether they also use the alternative exon 0 could be established only for one amphibian for which a transcriptome is available, the salamander Hynobius retardatus [27]. Analysis of this showed that exon 0 is present in this species with 24% exon 0-exon 2 splices and 76% exon1-ex2. A genomic sequence is not available for this species, precluding analysis of the sort shown in Fig. 2A. For this species, as well as in birds, in-frame ATG codons would allow translation to give alternative POU1F1 proteins differing only in the short N-terminal sequences encoded by exons 0 or 1 (Fig. 2C).

Fig. 2.

Alternative splicing at the 5′ end of the POU1F1 gene in birds. A. Expression profiles for exons 1–3 and the upstream region in 3 bird species, illustrating the presence of an additional exon (ex0) in chicken (Gallus gallus) and swan goose (Anser cygnoides), and two additional exons (ex0 and ex0b) in hooded crow (Corvus cornix). E, C and B indicate the positions of the 3 conserved enhancer sequences, with enhancer E located immediately upstream of ex0 in each case. Based on analysis of SRA experiments SRX3216799 (chicken), SRX4891147 (swan goose), SRX330897 (hooded crow). B. Alternative splice patterns seen in the above 3 species. Percentages at the right are the proportions of the splice variants derived from the SRA experiments referred to above. C. Amino acid sequences (derived by conceptual translation) for exon 0 spliced to exon 2 for the 3 bird species indicated above and a salamander (Hynobius retardatus). Potential translation start sites (methionine) are highlighted.

3.3. POU1F1 in fish

In teleost fish, identification of enhancer elements equivalent to those found in tetrapods initially proved difficult. However, elements B, C and E, but not A and D, were found in several other fish groups by Blast searches, and eventually were identified in many teleosts also. POU1F1 was identified in all fish groups examined except Agnatha (lamprey and hagfish). Elements B, C and E were found upstream of this in most species examined, including lungfish, sturgeon, gar, many teleosts and all 10 Chondrichthyes represented in the NCBI wgs database (Fig. 3). Sequence similarity between tetrapods and fish is clear for each of elements B, C and E (Supplementary Fig. S3); in every case the order E-C-B (5′ to 3′) is retained, though relative spacing varies greatly. This last is most notable in the case of lungfish which has a very large genome [28] and a correspondingly large POU1F1 gene (exon 1-exon 7–541 kb) and upstream region (element E-exon 1–523 kb; c.f. ∼5–38 kb in other fish). Element B was not found in coelacanth, probably due to a large unsequenced region between element C and exon 1.

Fig. 3.

Organization of the POU1F1 gene in fish. Black and grey bars represent the 7 exons. Red bars indicate conserved sequences (E,C,B) corresponding to the putative enhancers. Blue bars represent alternatively spliced upstream exon(s). In some cases where these are not shown (HemWyc, OruLat, GasAcu) analysis of transcriptomic data suggests that alternative upstream splicing does not occur, in others (LatCha, AciRut, LepOcu, AngAng, CalMil) presence or absence of upstream splicing cannot be evaluated owing to lack of appropriate transcriptomic data. Sequences included are: MusMus, mouse, LatCha, coelacanth, AciRut, sturgeon, LepOcu, spotted gar, AngAng, eel, DanRer, zebrafish, HemWyc, catfish, OncKis, salmon, OreNil, tilapia, OryLat, medaka, GasAcu, stickleback, CalMil, elephant shark, ChiPun, bamboo shark. For full species names see Supplementary Table S1. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

In some fish POU1F1 contains an additional, untranslated exon (exon 0) closely downstream of enhancer element E, similar to that seen in birds. In zebrafish and some close relatives, exon 0 is found in almost all transcripts; in others, for example salmon, it is found in a relatively small proportion (e.g. ∼6% in coho salmon, Oncorhynchus kisutch), with most transcripts initiating in exon 1.

3.4. POU1F1 in mammals

Analysis of the mouse pou1f1 gene showed the presence of 5 upstream enhancer elements, but the corresponding region of human revealed only four, with element D missing (Fig. 1). To assess the extent of such variability in mammals we examined the POU1F1 gene in a wide variety of mammals, including almost all those for which genome sequences are available in the wgs database (∼300 species). In all of these a single POU1F1 gene was identified, although in a few cases the sequence was incomplete. The 5-enhancer arrangement was found in at least some representatives of each of the main mammalian groups (Marsupialia, Afrotheria, Xenarthra, Laurasiatheria and Euarchontoglires) except Monotremata (Fig. 4; Supplementary Fig. S4). Enhancer elements A and C were detected in all mammals examined. Element B was present in all except elephant shrew (Elephantulus edwardii; Afrotheria) and gundi (Ctenodactylus gundi; Rodentia). On the other hand, elements D and E were absent from a number of groups. The organization of the regulatory region in a selection of mammals is shown in Fig. 4.

Fig. 4.

Organization of the POU1F1 gene in mammals. Black bars represent the 6 exons. Red bars indicate conserved sequences (A–E) corresponding to the accessible regions in the ATAC profile (top panel). A vertical arrow indicates site of a deletion removing the E element. Sequences included are: MusMus, mouse, RatNor, rat, NanGal, blind mole rat, CavPor, guinea pig, UroPar, arctic ground squirrel, OryCun, rabbit, HomSap, human, MacMul, macaque, PapAnu, baboon, CalJac, marmoset, MicMur, mouse lemur, CanLup, dog, SusScr, pig, TurTru, dolphin, MyoLuc, little brown bat, ManJav, pangolin, DasNov, armadillo, LoxAfr, elephant, MonDom, oppossum, OrnAna, platypus, TacAcu, echidna. For full species names see Supplementary Table S1. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Element D is not present in the sequence upstream of POU1F1 in monotremes (platypus and echidna). Since Monotremata is the outgroup for all other mammals this absence could be because monotremes represent an intermediate between the 3-element arrangement seen in most non-mammalian vertebrates and the 5-element structure of many mammals. On the other hand it is possible that the 5-element organization appeared early in mammalian evolution, and that element D was subsequently lost in Monotremata. Notably, the distance between elements C and E is considerably less in monotremes (∼3.8 kb) than in most other mammals (e.g mouse, 10 kb). Element D is also absent in all Primates and the two orders most closely related to primates, Scandentia and Dermoptera. This is clearly a consequence of loss of this element on this branch of Euarchontoglires, after separation from Glires (rodents and lagomorphs). Whether this was due to loss of element D by deletion or sequence divergence is not clear from the sequence data, given that it occurred at a fairly early stage of eutherian evolution. Element D was also absent in elephant shrew and two species in the rodent family Heteromyidae, pocket mouse (Perognathus) and kangaroo rat (Dipodomys).

Element E is present and well conserved in monotremes and marsupials. Indeed, in some marsupials, but not Monodelphis (opossum), it is duplicated. However, sequence conservation of Element E is relatively poor in Eutheria, and the element is missing in a number of groups. In most cases, this is clearly due to deletion. Thus in primates Element E is absent in all New World Monkeys (NWM) and in the group of Old World Monkeys (OWM) including Macaca and Colobus. Analysis of sequences shows that this is due to distinct, independent deletions. Independent deletions also removed element E in some rodents (Gliridae - dormice), all lagomorphs (rabbits and hares), some Carnivora (Canidae - dog family; sequence divergence high in other Carnivora), all Chiroptera (bats), several Eulipotyphla (moles and hedgehogs) and one Afrotherian (aardvark, Orycteropus). The lowered sequence conservation of element E suggests decreased importance in Eutheria, and the repeated deletions suggest that it may have been deleterious in some groups. Notably, deletion of element E in NWM and some OWM, in addition to the loss of element D in all primates, means that these groups retain only 3 enhancer elements, C, B and A. The hedgehog, Erinaceus, also retains only these 3 elements, while elephant shrew only has elements A, C and E.

Of note, an additional conserved sequence (element F) was identified upstream of element E in most mammals, but not in other vertebrates (Fig. 1, Fig. 4). This element was not associated with an ATAC-Seq peak and was upstream of the conserved CTCF element. There is no evidence that it is involved in regulating the POU1F1 gene, but such a role is possible. Future studies are needed to establish such possibility.

3.5. CTCF

A conserved CTCF binding site [29] was found upstream of the 5′-most enhancer element (element E except for those mammals in which this element is not present) in all tetrapods and in coelacanth and lungfish. This site is strongly conserved in Eutheria, but not monotremes or marsupials. An equivalent CTCF binding sequence was not found consistently in other fish. A conserved CTCF binding site is also found in the 3′-most intron in mammals, reptiles, birds, coelacanth and lungfish (Figs. 1 and 4). Alignments are shown in Supplementary Figs. S1 and S4.

3.6. Promoter

The promoter region upstream of exon 1 of POU1F1 has been characterized previously in rat [7,9] and shown to contain binding sites for POU1F1 itself, upstream (stimulatory) and downstream (inhibitory) of the transcription start site [9,30]. The POU1F1 sites are strongly conserved in mammals and indeed in all tetrapods, but less so in fish (Supplementary Figs. S1, S3 and S4). The mouse promoter contains a TATA box sequence, which is strongly conserved, with the extended sequence TATAAATAC being found in most mammals, all birds and amphibia, all non-teleost fish and most teleosts (Supplementary Figs. S1, S3-S5). Minor substitutions, unlikely to affect function occur in reptiles and some mammals and teleosts. Thus in dog (TATAAGCGC) and pig (CATAATAC) POU1F1 expression is as high as in mouse, rat and human (all TATAAATAC) (Supplementary Table S2). Significant substitutions, likely to abrogate function are seen in a few teleosts, including zebrafish, where obligate use of upstream exon 0 would imply that the promoter in front of exon 1 is less or no longer functional.

3.7. Expression levels

To determine whether changes seen in the organization of the POU1F1 upstream regulatory region during vertebrate evolution were associated with changes in overall expression of the gene, transcription in the anterior pituitary was assessed for a variety of species using transcriptomic data available from the NCBI Sequence Read Archive (SRA). Such data uses material from a wide range of studies, with animals of either sex and varying physiological states and subject to various experimental conditions. Results are shown in Supplementary Table S2. In all cases studied, expression levels were high, and there was no clear evidence for gross changes in expression level associated with varying organization of the regulatory region of the gene. This does not rule out more-subtle changes, for example reflecting physiological or developmental state.

3.8. DNA accessibility in non-mammals

Identification of additional potential regulatory elements upstream of POU1F1 was initially based on the DNA accessibility profile in mouse (Fig. 1). The observation that accessibility is associated with sequence conservation was then used to identify sequence regions potentially involved in gene regulation in other species. In order to support this involvement, DNA accessibility studies were carried out on several non-mammalian species. Using the established protocol of ATAC-seq [16,31] we have assessed DNA accessibility in guinea fowl, alligator and bullfrog. In brief, we have dissected pituitaries of each animal, isolated nuclei and performed ATAC-seq according to the well established laboratory protocol [16,32]. Two animals per group were analyzed. ATAC-seq raw reads data were mapped to specific genome and visualized using IGV software (Fig. 5A) [33,34]. Data were presented together with schematic representation of the region for each animal and conserved elements. Our analysis demonstrated that the conserved regions in these species correspond to regions of high DNA accessibility in pituitaries, suggestive of functionally active regulatory elements.

Fig. 5.

DNA accessibility upstream of the POU1F1 gene in selected non-mammalian species, studied using ATAC. A. Examples of DNA accessibility in three species as compared to the mouse regulatory region (top). Alignment of open regions in alligator (AllMis), guinea fowl (NumMel) and bullfrog (LitCat) shows accessible DNA in regions corresponding to the conserved elements as found by the sequence similarities. B. DNA accessibility in the goose (Anser cygnoides, AnsCyg) [35] regulatory region showing conservation of the regions and their DNA availability for regulation.

Similarly, we used recently published goose pituitary ATAC-seq data [35] to align the conserved sequences with the DNA accessibility of goose POU1F1 regulatory region. As in the case of guinea fowl, alligator and bullfrog, highly conserved regulatory elements were open in the goose pituitary, strongly suggesting the functionality of these regions (Fig. 5B).

3.9. Enhancer alignments

Alignments for the five enhancer elements are shown in Supplementary Figs. S1, S3 and S4. In most cases, there is ‘structure’ within the alignment, with regions of relatively high similarity separated by poorly conserved regions. This is particularly marked in the case of element B where 4 such subregions can be seen. In some fish, including medaka and stickleback, these are more widely dispersed, and less distinct. Notably this structuring and spacing is retained in axolotl and lungfish, despite the drastically expanded genome in these species, with distances between enhancer elements increased almost 10-fold compared with other species.

Autoregulatory POU1F1 sites have been previously identified in the POU1F1 promoter and enhancer element C in rodents [9,12]. Many of these sites are strongly conserved across tetrapods (Supplementary Figs. S1, S5). Thus, of the 5 POU1F1 binding sites identified on enhancer DE in mouse [13], three are identifiable and conserved in the element C alignment (Supplementary Figs. S1, S5), but the other two are upstream of this in poorly conserved sequence. The two POU1F1 sites identified in the promoter [9] are well conserved. Potential POU1F1 binding sites were identified in element E, where they are strongly conserved in non-mammalian tetrapods, but not in placental mammals (Supplementary Fig. S5). The mammalian-specific elements A and D also contain potential POU1F1 binding sites, with up to three in element A, though these are poorly conserved (Supplementary Fig. S5). The presence of strongly conserved POU1F1 binding sites in element E in non-mammals that are much less well conserved in placental mammals, suggests that autoregulation of this element by POU1F1 may be important while it functions as a promoter, but less so when this role is lost. Appearance of POU1F1 binding sites in elements D and A, unique to mammals, would allow the overall autoregulatory role to be maintained and perhaps strengthened in mammals.

4. Discussion

4.1. Regulatory elements of the POU1F1 gene

Previous studies have established that expression of the POU1F1 gene in mouse and rat is controlled by a number of transcription factors, including POU1F1 itself, via interactions with its promoter and at least three upstream enhancer sequences [6,7,[9], [10], [11], [12], [13],24]. Here we propose that in mouse there are 5 potential regulatory elements (enhancers) upstream of the mouse POU1F1 promoter, on the basis of DNA accessibility and sequence conservation. These are designated elements E-A (5′ to 3′). They are flanked by CTCF binding sites upstream of element E and in intron 5 of the POU1F1 gene, suggesting that they may constitute a cluster of enhancers cooperatively regulating expression of the locus [29]. Element C corresponds to the previously-described definitive enhancer (DE; 11) while elements A and B correspond to the early enhancers (EEα, EEβ, respectively; [11,13,24]).

In the study reported here, the nature of the regulatory elements associated with POU1F1 was explored by examining genome sequences across vertebrate groups. The POU1F1 gene was identified in all groups except the Agnatha (lamprey and hagfish). In all these groups, conserved sequences corresponding to enhancer elements B, C and E were detected, but elements A and D were only detected in mammals (Fig. 6). Notably, in many non-mammalian vertebrates for which pituitary transcriptomes are available (birds, salamander, teleosts and chondrichthyes) transcription can start at an upstream site, with use of an alternate exon (exon 0) which is spliced into exon 2 (birds, salamander, Fig. 2B) or exon 1 (fish). In most cases transcripts starting at exon 0 are no more abundant than those starting at exon 1 (and in some fish, transcription from exon 0 was not apparent), but in zebrafish exon 0 appears to be used exclusively. In all cases exon 0 lies closely downstream of element E, suggesting that the latter acts here as an alternative promoter. DNA accessibility studies supported the existence of the regulatory elements B, C and E in a representative bird, reptile and amphibian. There was no suggestion of use of exon 0 or another alternative upstream exon and transcription start site in placental mammals for which pituitary transcriptomes are available (several species of rodents, primates, carnivores and artiodactyls). An alternative transcription start site at the 5’ end of a gene, as seen here, can have two consequences. First it can lead to production of a protein with an alternative sequence at the N-terminus. In the case of POU1F1 this alternative coding sequence is short, and poorly conserved, suggesting lack of a specific function (Fig. 2C). Second, it can lead to production of transcripts under an alternative regulatory mechanism by use of an alternative promoter and potentially different enhancer elements. This seems more likely, given the conservation of the putative promoter, and its equivalence to element E.

The studies described in this paper have involved primarily investigation of conserved sequences upstream of the POU1F1 locus in a wide range of vertebrate groups and, in several species, of the accessibility of such sequences as an indicator of possible function. Interpretation of the functional significance of the results is based on previous experimental studies, carried out mainly in rodents (rat and mouse). To obtain a fuller understanding of the evolution of POU1F1 across vertebrates, functional studies on a much wider range of species, including both other mammalian groups and non-mammalian vertebrates are needed. These could include luciferase-based assays to determine the activities of individual potential enhancer sequences, as used previously for rodents [[11], [12], [13]] and/or targeted deletion of such sequences to assess in vivo function. Comparative study of the transcription factors, and their associated co-factors, involved in regulation of POU1F1 expression would be of interest, though their identification in rodents is still incomplete. Technologies such as chromatin immunoprecipitation sequencing and three-dimensional chromosome conformation studies should be applied to map transcription factor binding sites and their organization across vertebrates. Application of single-cell RNA sequencing to mature and/or developing anterior pituitary could help determine whether regulation of POU1F1 expression varies between cell types, in rodents and other species.

4.2. Evolution of the regulatory region

Our observations suggest that the regulatory mechanisms for the POU1F1 gene have been maintained fairly constant across all non-mammalian vertebrates except Agnatha, with use of alternative promoters playing an important role (Fig. 6A). With the appearance of mammals substantial changes occurred, with the acquisition of new regulatory elements (elements A and D) and loss of the alternative exon associated with element E, the latter potentially acquiring function as an enhancer. Element A must have been acquired early - it is found in all mammalian groups, but element D is not present in monotremes, so could have been acquired later.

Exactly when exon 0 was lost is less clear. It is present in all bird species examined. Sequence similarity (based on Blast analysis) suggests that it may also be present in reptiles, amphibia, monotremes and marsupials, and presence in amphibia is supported by data from the pituitary transcriptome of a salamander. Across mammals, the composition of the POU1F1 regulatory region appears to be less stable/conserved than is seen for other vertebrates. Elements A, B and C are found in nearly all mammalian groups, but element D is absent in primates and the two groups most closely related to primates (tree shrews and flying lemurs), while element E is absent from a number of groups, reflecting at least 6 independent deletions (Fig. 6B). This last suggests that element E, having lost its role as a promoter for exon 0, may play a rather equivocal role in mammals; where present its sequence is usually conserved, suggesting retention of specific function, but complete removal by deletion on so many separate occasions suggests that its presence may even have been disadvantageous in some cases. It is notable that for at least two primate groups (NWM and some OWM, including macaques) deletion of element E and absence of element D, leaves just elements A, B and C, with no apparent effect on level of POU1F1 expression. Notably, the variability of the regulatory region of the POU1F1 gene in mammals contrasts with the strongly conserved nature of the coding sequence [36].

The most important change occurring during the evolution of POU1F1 regulation appears to be that associated with the appearance of mammals, with loss of exon 0 and the associated alternative promoter, and gain of two additional enhancer elements, A and D. However, a number of additional features are also worth noting, including the switch in orientation of element C seen in Anura (frogs and toads) and retention of an apparently normal 3-element organisation in axolotl and lungfish, despite (independent) expansions of genome size with a ten-fold increase in distance between element E and promoter.

The origins of the 5 enhancer elements are unclear. Elements B, C and E are clearly present in Chondrichthyes, and are well conserved across almost all vertebrate groups but, like the POU1F1 gene, could not be detected in Agnatha. Sequence comparisons of the enhancer elements detected no similarity between them. The origin of the additional elements (A and D) seen in mammals is not clear. Element D is not present in monotremes; it may have appeared later in mammalian evolution or may have been present in the ancestor of monotremes and theria (placentals and marsupials), and subsequently lost in the former. The most interesting case is that of element E. Our data show that in birds and some amphibia and fish element E is immediately upstream of the additional 5′ exon (exon 0). It is possible that element E was originally an upstream enhancer that was taken over as a promoter in some groups. It is also possible that transcription from exon 0, under the control of the alternative promoter (Element E), allows separate regulation of POU1F1 expression in early development in non-mammalian vertebrates, this function being taken over by the early enhancer EEα (Element A) in mammals. Another possibility is that element E originated as an alternative promoter. In this case, during the co-transcriptional splicing, element E would be in close proximity to the main promoter which would increase local concentration of transcriptional cofactors and therefore enhanced transcription from the main promoter – a role, as we understand it today, attributed to enhancers. As such, our data, may present a novel evolutionary mechanism for the origin of some enhancers.

Element A is uniquely present and strongly conserved in all mammals. It includes the EEα enhancer that has been shown experimentally to be required for early expression of POU1F1 [11,13,24], although the conserved region of element A is much bigger than EEα. Extensive Blast searches of mammalian and non-mammalian genomic and other sequence databases did not reveal any sequence clearly related to element A other than upstream of the POU1F1 gene in mammals. In early developing pituitary, element A is occupied by the giant platform-like protein ATBF1, that through its multiple DNA- and protein-binding domains consolidates multiple signals regulating the early expression of the gene [13], before POU1F1 transcription factor takes over its own regulation in autoregulatory fashion. ATBF1 is expressed only during a very short window of pituitary and brain development and only in a very limited number of tissues, raising the possibility that its role in the pituitary is to keep the POU1F1 regulatory region available for regulation until appearance of POU1F1 to take over the transcriptional regulation of its locus.

The availability of genomic sequences from several hundred vertebrate species provides an abundance of data for comparative studies of the type presented here. The approach is inevitably broad-brush, partly because the data are sometimes incomplete and also because of intrinsic limitations. The main novel observation made in this study concerns the switch which occurred with the appearance of mammals, with the acquisition of additional elements in the POU1F1 regulatory region, potentially giving rise to a super-enhancer, along with loss of the alternative upstream exon and promoter. The switch seems to involve a change from use of an alternative promoter/exon as a significant feature in regulation of POU1F1 transcription to regulation based solely on a number of enhancers. Such a change seems to be substantial, but it may in part reflect the rather close relationship between promoters and enhancers which is now well recognized [1,5].

The functional changes associated with this switch are not clear. The overall role of POU1F1 seems similar in most vertebrates, and transcription levels in the pituitary are high throughout. A biological function specific to mammals could involve the regulation of prolactin with its role in controlling lactation. In the female mammal, prolactin secretion undergoes large variations associated with pregnancy, lactation and post-lactation involution of the mammary gland. These changes may involve recruitment of additional lactotrophs in preparation for lactation, and subsequent reversal of the process - involving substantial cellular plasticity [37]. Involvement of POU1F1 in these processes could require transcriptional controls additional to those needed for development of the pituitary gland and maintenance of high expression of growth hormone, prolactin and thyrotropin. This need may have been met by acquisition of additional enhancer elements (A and D) with subsequent modification involving loss of the alternative upstream exon and in some cases elements E and/or D. The actions of prolactin on the mammary gland are of course confined to mammals; it has many other functions in lower vertebrates, but little is known about its control by POU1F1 there. It is also possible that acquisition of the additional strong regulatory element (A) in the region open early during the development of the pituitary was necessary to prevent heterochromatinization of the locus during brain development in mammals where increased complexity of the genome required a safeguarding mechanism for genes indispensable for proper function of the organism.

In sum, in this work we have followed the evolution of the regulatory region of the POU1F1 gene and found a potential sequence of events that led to the self regulatory, super-enhancer-like region maintaining high and reliable expression of the gene in the adult mammalian pituitary. Our data not only suggest how an alternative promoter might become an enhancer but also underline the potential importance of maintenance of old regulatory elements in parallel with introduction of new elements in order to maintain gene expression and function.

Data availability

Sequencing data and other information will be shared upon request.

CRediT authorship contribution statement

Michael Wallis: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Funding acquisition, Formal analysis, Data curation. Qianlan Xu: Writing – original draft, Visualization, Resources, Methodology, Investigation, Formal analysis, Data curation. Michal Krawczyk: Writing – original draft, Visualization, Investigation, Formal analysis, Data curation. Dorota Skowronska-Krawczyk: Writing – review & editing, Writing – original draft, Supervision, Resources, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: