Cis-Acting Transcriptional Repression Establishes a Sharp Boundary in Chordate Embryos

See allHide authors and affiliations

Science  24 Aug 2012:
Vol. 337, Issue 6097, pp. 964-967
DOI: 10.1126/science.1222488


The function of bone morphogenetic protein (BMP) signaling in dorsoventral (DV) patterning of animal embryos is conserved among Bilateria. In vertebrates, the BMP ligand antidorsalizing morphogenetic protein (Admp) is expressed dorsally and moves to the opposite side to specify the ventral fate. Here, we show that Pinhead is an antagonist specific for Admp with a role in establishing the DV axis of the trunk epidermis in embryos of the ascidian Ciona intestinalis. Pinhead and Admp exist in tandem in the genomes of various animals from arthropods to vertebrates. This genomic configuration is important for mutually exclusive expression of these genes, because Pinhead transcription directly disturbs the action of the Admp enhancer. Our data suggest that this dual negative regulatory mechanism is widely conserved in animals.

In many animals, including Xenopus, zebrafish, and planarians, specification of the dorsoventral (DV) axis depends on the spatially opposed expression of Bmp4/Bmp7 and Admp (17). Although a basal chordate Ciona intestinalis lacks a structure homologous to the vertebrate organizer (8), which coordinates the formation of mesoendoderm and the induction of neural tissues, Admp plays a key role in specification of the DV axis in the tail ectoderm of the larvae (9, 10). In addition, the expression pattern of Admp, Bmp2/4, and Chordin, which encodes a bone morphogenetic protein (BMP) antagonist, along the DV axis of the trunk epidermis in the tailbud-stage embryo is reminiscent of that in vertebrate gastrulating embryos (11).

To better understand Bmp signaling, we searched the Ciona genome for genes encoding a cysteine-knot domain, which is common to all transforming growth factor–β superfamily ligands and many of their antagonists (table S1). A homolog of Xenopus Pinhead (12) was of particular interest, because its genomic position is next to Admp.

Ciona Pinhead encodes a protein with three putative cysteine knots (fig. S1A). Initially, Pinhead is weakly and transiently expressed in the early embryo (fig. S1B). Strong continuous expression occurs in the posterior ventral epidermal cells at the gastrula stage. This epidermal expression expands anteriorally to the entire ventral epidermis up to the neurula stage. At the tailbud stage, the ventral epidermal expression is restricted to the trunk region (arrow in Fig. 1O), and is also expressed in the ventral endoderm (arrowhead in Fig. 1O). Pinhead expression overlaps with Bmp2/4 expression (Fig. 1L) but does not overlap with Admp expression, which is observed in the lateral epidermis and in the dorsal endoderm (11) (Fig. 1I).

Fig. 1

Expression of (A to D) NK4, which marks the ventral epidermis; (E to H) FoxB; (I to K) Admp; (L to N) Bmp2/4; and (O to Q) Pinhead in (A), (E), (I), (L), and (O) control embryos and morphant embryos injected with [(B), (F), (M), and (P)] Admp, [(C), (G), (J), and (Q)] Bmp2/4, and [(D), (H), (K), and (N)] Pinhead MOs. The scale bar in (A) represents 100 μm. Inserts in (I) and (L) to (Q) are in ventral views of the trunk region.

We injected morpholino antisense oligonucleotides (MOs) against Pinhead, Admp, or Bmp2/4 into fertilized eggs and examined their effects at the tailbud stage. NK4 expression, which marks the ventral epidermis in the trunk, was suppressed in Admp morphants (Fig. 1, A and B). FoxB expression, which marks the lateral epidermal region, was expanded into the ventral region (Fig. 1, E and F). In Bmp2/4 morphants, Nk4 expression was reduced, whereas we could not observe distinct change of FoxB expression (Fig. 1, C and G). In Pinhead morphants, NK4 expression was expanded to the lateral regions, and FoxB expression was reduced (Fig. 1, D and H). Thus, BMP signaling is required for ventral fate and is negatively regulated by Pinhead in the trunk epidermis.

Admp expression appeared normal in Bmp2/4 morphants (Fig. 1J) but was narrowed in Pinhead morphants (Fig. 1K). Bmp2/4 expression was suppressed in Admp morphants (Fig. 1M) and expanded in Pinhead morphants (Fig. 1N). Pinhead expression was suppressed in Admp and Bmp2/4 morphants (Fig. 1, P and Q) (see also supplementary text and fig. S2). Thus, the expression of these three genes is interrelated.

Next, we investigated whether Pinhead interacts directly with Bmp and/or Admp. Constructs, in which 3x Flag-tagged Pinhead and 3x Myc-tagged Admp (or BMP2/4) were expressed in the entire epidermis under the Dll-b upstream sequence, were coelectroporated into fertilized eggs. Flag-tagged Pinhead coimmunoprecipitated strongly with Myc-tagged Admp in embryonic lysates (Fig. 2) and vice versa (fig. S3). By contrast, only a very weak interaction was observed between Flag-tagged Pinhead and Myc-tagged BMP2/4.

Fig. 2

Lysates of embryos with overexpression of Flag-tagged Pinhead and Myc-tagged Admp (or Bmp2/4) were immunoprecipitated (IP) with antibodies to Myc. The bounded proteins were analyzed by Western blotting (WB) with antibodies to Flag.

Previous studies (7, 13) have identified Admp orthologs in the genomes of lophotrochozoans and various deuterostomes, but not in mammals. We also found Admp in the genomes of ecdysozoans, such as the water flea Daphnia pulex, the wasp Nasonia vitripennis, and the honeybee Apis mellifera (fig. S4). Pinhead is present in the genomes of Daphnia, Nasonia, Apis, sea urchin, amphioxus, fugu, zebrafish, and Xenopus (fig. S5). In all of the genomes in which a Pinhead gene was found, it was upstream of and oriented in the same direction as Admp.

To examine the cis-regulatory mechanisms of these two genes, we cloned 7-kb genomic sequences from –1.0 kb upstream of Pinhead to the 3′ end of Admp gene and replaced Pinhead and Admp with red fluorescent protein (RFP) and green fluorescent protein (GFP) reporter genes. Because Admp is transiently expressed under the control of Dll-b in all the epidermal cells at the early gastrula stage (fig. S6), perdurance of GFP might be observed after transcription is terminated. Therefore, we deleted a region that contains a cluster of six putative homeobox binding sites responsible for the expression of Admp at the gastrula stage [–51 to –267 base pairs (bp) upstream of the 5′-most transcriptional start site (TSS) of Admp]. As a result, this RFP/GFP reporter construct recapitulates the endogenous expression of Pinhead and Admp in the tailbud-stage embryo (Fig. 3, A and A′). The Pinhead upstream sequence alone also reproduced its endogenous expression, although the expression was weaker (Fig. 3B). In contrast, with the Admp upstream sequence alone, ectopic reporter expression was observed in the ventral region in addition to the lateral expression (Fig. 3C).

Fig. 3

Analysis of the regulatory regions of Admp and Pinhead. (A to I) The illustrations on the left depict the constructs used. The red boxes indicate the RFP gene and SV40 polyadenylation signal that replaced Pinhead, and the green boxes indicate the GFP gene and SV40 polyadenylation signal that replaced Admp. The numbers indicate the relative nucleotide positions from the 5′-most TSSs of Pinhead and Admp. The blue ovals indicate putative SBEs, and the yellow ovals indicate E boxes important for Admp expression. Mutated binding sites are indicated by X. The two TSSs (3 bp apart from each other) of Pinhead and the four TSSs (distributed within a 356-bp region) of Admp are indicated by arrows in (A). Pink and light green boxes indicate the Pinhead promoter region and the Admp promoter region, respectively. Red and green bars in the middle column indicate scoring of RFP and GFP expression in the ventral epidermis, respectively. Red and green bars in the right column indicate scoring of RFP and GFP expression in the lateral epidermis, respectively. (A′, C′, H′, and I′) Merged images showing RFP and GFP expression in embryos injected with the constructs shown in (A), (C), (H), and (I), respectively. Green- and red-channel images of the trunk region are shown in (A′), (H′), and (I′).

A series of deletions of the Admp upstream sequence revealed that the region between –267 and –650 bp enhanced RFP expression (Fig. 3D and fig. S7, A to C). We hereafter call this region the G enhancer. Similarly, the region between –650 and –972 bp was essential for GFP expression in tailbud-stage embryos (fig. S7, D to F), whereas the G enhancer enhanced it (Fig. 3D). There are three E boxes between –650 and –972 bp conserved in the closely related Ciona savignyi. Mutations in these E boxes completely abolished GFP expression (Fig. 3E). We call the region containing these E boxes the A enhancer.

The expression of RFP gradually decreased in a series of deletions of the Pinhead upstream region, although a construct with the 383-bp upstream sequence exhibited a similar level of RFP expression in the ventral region compared with the full-length construct (fig. S7, G to J). Because Pinhead is controlled by Admp/Bmp signal, we mutated four Smad-binding elements (SBEs) within this 383-bp region. This mutated construct did not drive RFP expression (Fig. 3F). We call the region containing these SBEs the P enhancer.

In the Pinhead-upstream sequence series of deletions, we observed increased ectopic GFP expression in the ventral region concomitant with decreased RFP expression (fig. S7, G to J). Mutation of the SBEs also resulted in ectopic GFP expression in the ventral region (Fig. 3F). However, the Pinhead upstream region tethered directly to the Admp upstream sequence did not repress GFP expression in the ventral region (Fig. 3G). Therefore, the P enhancer does not directly repress GFP expression in the ventral region. Instead, these observations indicate that Pinhead transcription interferes with Admp transcription (see also supplementary text and fig. S7, K to M).

This transcriptional link was conserved in medaka. A construct with the upstream sequences of medaka Pinhead and Admp, driving RFP and GFP, respectively, caused nonoverlapping expression of GFP and RFP in the axial tissues and the surrounding tissues of 2-day embryos in most cases (fig. S8A) (58%, n = 29 embryos) and barely drove GFP expression in the surrounding tissues (7%). In contrast, a construct containing only the Admp upstream region and GFP drove GFP expression in both of the axial tissues and the surrounding tissues (fig. S8, B and B′) (42%, n = 47 embryos).

The above results suggested that the repression of Admp transcription in the ventral region is due to a cis-acting mechanism associated with Pinhead transcription. Transcriptional interference by the read-through transcript is not likely the principal mechanism, because we rarely detected a considerable amount of transcript around the 3′-most E box in the A enhancer (fig. S9), which alone can activate Admp transcription (fig. S7D).

Next, we examined chromatin structure by the chromosome conformation capture (3C) method (14) using two types of embryos. In the first type, Bmp2/4 was overexpressed under the Dll-b enhancer, resulting in down-regulation of Admp and up-regulation of Pinhead (fig. S10, A and B). In the second type, noggin was overexpressed, resulting in up-regulation of Admp and down-regulation of Pinhead (fig. S10, C and D). A common primer on fragment 5, which contains the three SBEs and the Pinhead promoter, was used in combination with a set of primers along the genomic loci. A specific and strong interaction with fragment 10 containing the A enhancer and most of the G enhancer was observed in embryos with up-regulated Pinhead (Fig. 4A). This was confirmed with another common primer on fragment 10 (fig. S10E). Thus, when Pinhead is transcribed, the Pinhead upstream region interacts with the region containing the A and G enhancers (Fig. 4B).

Fig. 4

(A) 3C analysis examining interaction of the Admp enhancer with nearby regions. A schematic representation of the genomic organization of Pinhead and Admp and restriction fragments used is shown below the graph. Blue and yellow triangles indicate the SBEs and E boxes in the P and A enhancers. The brown box indicates the G enhancer. The graph shows interactions of fragment 5 with the other fragments. The y axis indicates normalized interaction frequency calculated as fold difference relative to the control reaction using bacterial artificial chromosome DNA. Blue bars indicate embryos with Bmp2/4 overexpression, which leads to up-regulation of Pinhead transcription. Yellow bars indicate embryos with noggin overexpression, which leads to down-regulation of Pinhead transcription. Error bars indicate standard errors of quantitative real-time fluorescence polymerase chain reaction. (B) Models of chromosome conformations of the P (blue), A (yellow), and G (brown) enhancers when Pinhead transcription is active (left) and inactive (right).

This interaction likely occurs in the Pinhead promoter but not in the P enhancer, because derepression of GFP in the ventral region was observed with a reporter construct lacking the Pinhead promoter (fig. S7N). Furthermore, replacement of the Pinhead promoter with the Admp promoter resulted in derepression of GFP in the ventral region despite the expression of RFP (Fig. 3, H and H′).

Because RFP was also expressed ectopically in the lateral region with the construct shown in Fig. 3H, the comparison between Fig. 3A and H suggested that the A enhancer acts only on the Admp promoter, not on the Pinhead promoter. Indeed, GFP was not driven with a construct in which the Admp promoter was replaced with the Pinhead promoter (fig. S7O). In addition, as described earlier, the G enhancer, but not the A enhancer, enhances Pinhead transcription (e.g., Fig. 3, D and E). These observations suggested that the G enhancer interacts with the Pinhead promoter, and the A enhancer is thereby sequestered, when Pinhead is transcribed. To investigate this possibility, we swapped the A and G enhancers. This swap construct, in which the interaction between the G enhancer and the Pinhead promoter could not sequester the A enhancer, indeed resulted in derepression of GFP in the ventral region (Fig. 3, I and I′). Therefore, we conclude that the interaction between the G enhancer and the Pinhead promoter disturbs the action of the A enhancer (see also supplementary text).

Bmp signaling is widely used in determining the DV axis of bilaterian embryos. Our data indicate that Pinhead plays a key role in this system (fig. S11). Because Pinhead specifically interacts with Admp, Pinhead antagonizes Admp activity after Admp induces Pinhead transcription, preventing Admp activity from propagating further laterally and dorsally, but does not affect Bmp2/4 signaling. Furthermore, Pinhead transcription suppresses Admp transcription, thereby ensuring their mutually exclusive expression. This occurs mainly through competition for the G enhancer, a competition that Pinhead always wins when the Admp/BMP signal is active. Because the genomic configuration of Pinhead and Admp is widely conserved from arthropods to vertebrates, this dual negative regulation may be conserved as well.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S11

Tables S1 to S3

References (1529)

References and Notes

  1. Acknowledgments: This research was supported by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology and the Japan Society for the Promotion of Science (JSPS) to K.S.I. (22870018), T.G.K. (22310120), and Y.S. (21671004) and also in part by the Global Center of Excellence program (A06) to Kyoto University. K.S.I. was supported by a Restart Postdoctoral Fellowship (RPD) from JSPS. The cDNA sequences for Admp and Pinhead are available under accession nos. AB210298 and AK113761 in the DDBJ/EMBL/GenBank database. We are grateful to K. Hirayama for her assistance with this research.

Stay Connected to Science

Navigate This Article