Ventroptin: A BMP-4 Antagonist Expressed in a Double-Gradient Pattern in the Retina

See allHide authors and affiliations

Science  06 Jul 2001:
Vol. 293, Issue 5527, pp. 111-115
DOI: 10.1126/science.1058379


In the visual system, the establishment of the anteroposterior and dorsoventral axes in the retina and tectum during development is important for topographic retinotectal projection. We identified chick Ventroptin, an antagonist of bone morphogenetic protein 4 (BMP-4), which is mainly expressed in the ventral retina, not only with a ventral high–dorsal low gradient but also with a nasal high–temporal low gradient at later stages. Misexpression of Ventroptinaltered expression patterns of several topographic genes in the retina and projection of the retinal axons to the tectum along both axes. Thus, the topographic retinotectal projection appears to be specified by the double-gradient molecule Ventroptin along the two axes.

Axonal connection patterns in the nervous system often form topographic maps, with nearest neighbor relationships of the projection neurons maintained in their connections within the target. The projection from the retina to the tectum is a good model system for understanding the development of topographic maps. Graded distributions of topographic molecules along the anteroposterior (A-P) (nasotemporal) and dorsoventral (D-V) axes in the retina and tectum, which are derived from the regional specialization along the two axes during retinal and tectal development, control the topographical projection of retinal axons (1). Although gradients of diffusible factors and transcription factors are known to control the regional specificities in the retina during development (2–6), the molecular mechanisms involved are mostly unknown. We performed large-scale screening of region-specific molecules (7) [also see supplemental Web information (8)], using a new cDNA display system: restriction landmark cDNA scanning (RLCS) (9). We found a molecule that has activity to antagonize the function of bone morphogenetic protein 4 (BMP-4).

Figure 1A shows corresponding parts of a pair of RLCS profiles for the dorsal and ventral retina when Bam HI was used as the first restriction enzyme. The cDNA fragment corresponding to the ventral-specific spot shown with an arrow in Fig. 1A (tentatively named V/Bam HI 1) was recovered from the gel and was subcloned into a plasmid for further analysis. Northern blotting using this fragment as a probe revealed a single 4.9-kb transcript in the ventral retina at embryonic day 8 (E8) (Fig. 1B). This molecule was repeatedly identified not only as ventral-specific but as nasal-specific in our RLCS screening (8).

Figure 1

Isolation of Ventroptin by the RLCS method. (A) RLCS profiles of the dorsal and ventral retina. The V/Bam HI 1 (Ventroptin) spot is indicated by an arrow. The spot with an arrowhead was retinaldehyde dehydrogenase 1 (7). (B) Ventroptin mRNA expression in the dorsal (D) and ventral (V) retina. RNA blot hybridization analysis (10 μg of total RNA per lane) was done using the cDNA fragment of V/Bam HI 1 as a probe. The dorsal or ventral one-third of the retina was used for RNA preparation for RLCS and RNA blotting. GAPDH, glyceraldehyde phosphate dehydrogenase. (C) Alignment of chick (cVOPT) and mouse (mVOPT) Ventroptin sequences. The Genetics Computer Group program PileUp was used. The locations of the putative signal peptide (SP) and three CRs are shown. Amino acids that are identical between chick and mouse are shadowed in black; those that are similar are shadowed in gray. Dashes indicate gaps. Dots indicate amino acids that are identical between mouse Ventroptin-α and mouse Ventroptin-β. Amino acid identity and similarity between chick Ventroptin and mouse Ventroptin-α are 80 and 86%, respectively.

Full-length V/Bam HI 1 clones isolated from an E8 chick retina cDNA library encoded a protein of 456 amino acids, with a putative signal peptide and three cysteine-rich repeats (CRs) (Fig. 1C; GenBank accession no. AF257352). CRs are characteristic motifs that are conserved in some proteins, including von Willebrand factor and Chordin. Here we refer to this molecule as Ventroptin (VOPT), after the place where it is mainly expressed. Ventroptin is likely to be a secretory protein, because it was found in culture supernatant prepared from chick embryonic fibroblasts (CEFs) infected with a recombinant expression virus vector for Ventroptin (RCAS/VOPT-flag) (8) as a single band of about 50 kD by Western blot analysis (10). This size coincides with the predicted molecular weight of the Ventroptin protein when processed at the signal sequence (Fig. 1C).

We cloned mouse Ventroptin cDNA to deduce the structure of the mouse ortholog (Fig. 1C) (8). Two splicing isoforms were found for mouse Ventroptin: a long form (Ventroption-α; accession no. AF321853) and a short form (Ventroption-β; accession no. AF296451). The β isoform was devoid of the region after the third CR, as compared with the chick counterpart and the α isoform (Fig. 1C). Online database screening also identified a partial sequence of the human gene (accession no. AL049176).

In the chick eye, Ventroptin expression was first detected in the optic vesicle at Hamburger-Hamilton stage 11 by reverse transcription polymerase chain reaction (10) and became evident in the ventral retina at stage 14 when the optic vesicle invaginates, when examined by in situ hybridization (Fig. 2A). Ventroptin expression became still more manifest at E3 (stage 16 to 17) (Fig. 2B), when retinal neurogenesis just begins (11), seemingly in a complementary pattern to that of BMP-4 (Fig. 2C). Another BMP member, BMP-7, is ubiquitously expressed in the eye region (12). Figure 2D shows Ventroptinexpression in the retina with a ventral high–dorsal low gradient at E8. This D-V graded expression pattern was observed until E14, peaking at E8 (10). At E3, Ventroptin showed a uniform expression along the A-P axis. However, a nasal high–temporal low gradient expression pattern was detected first at E6 and became obvious at E8 (Fig. 2E). At this stage, Ventroptin showed a double-gradient expression profile along the two axes (Fig. 2, F and G). At E3, before the retinal lamination, Ventroptinexpression was observed throughout the ventral retinal tissue (Fig. 2B, inset). However, at E8, when retinal lamination becomes defined,Ventroptin-positive cells were localized to the inner nuclear layer (Fig. 2, D and E). This is evident in the central retina, where the development is advanced. Ventroptin was also expressed in organs other than the retina (13).

Figure 2

Expression patterns of VentroptinmRNA during development. (A) Coronal section in situ hybridization of the eye at stage 14. (B and C) Whole-mount in situ hybridization of E3 (stage 16 to 17) chick embryos for Ventroptin (B) and BMP-4 (C). Insets show results in coronal sections of the E3 eyes. Ventroptin (B) was expressed in the ventral retina (arrowheads) and the forebrain (arrow). BMP-4 (C) was expressed in the dorsal retina (arrowheads) and the periphery of the nasal pit (arrow). (D) Coronal section in situ hybridization at E8. Ventroptin is expressed in a ventral high–dorsal low (D ↔ V) gradient. (E) Horizontal section at E8. Ventroptin is expressed in a nasal high–temporal low (N ↔ T) gradient. (F) Flat-mount in situ hybridization of E8 retina. (G) Schematic drawing of Ventroptinexpression in the retina. Double-gradient expression is represented by the density gradient of color. Scale bars, 600 μm in (D) and (E). N, T, D, and V indicate nasal, temporal, dorsal, and ventral, respectively.

Several secretory proteins that antagonize BMPs have been discovered to date (14): Noggin, Chordin, Follistatin, Cerberus, and Gremlin. These BMP antagonists specifically bind to BMPs and prevent their binding to specific receptors or their signaling. Because the three CRs of Ventroptin were significantly homologous with those of Chordin, although the remaining region has no homology, we speculated that Ventroptin binds to BMPs. Binding between Ventroptin and several transforming growth factor–β (TGF-β) family members was tested with a surface plasmon resonance biosensor. Ventroptin bound with high affinity to BMP-4 and with lower affinity to a BMP-4/7 heterodimer, but not at all to BMP-7, TGF-β, or activin [Web fig. 1A (8)]. Furthermore, BMP-4 is coimmunoprecipitated with Ventroptin [Web fig. 1B (8)].

The binding of Ventroptin to BMP-4 suggested that Ventroptin can inhibit BMP functions as Chordin does. We injected earlyXenopus embryos with synthetic mRNA encoding chick Ventroptin [Web fig. 1C (8)]. Ventral expression ofVentroptin induced the formation of a secondary body axis in the injected embryos (in 66 out of 104 embryos). When BMP-4mRNA was coinjected, secondary axis formation induced byVentroptin mRNA was completely repressed (in 57 out of 57 embryos), indicating that Ventroptin can bind to BMP-4 and inhibit its activity in vivo.

The dorsal retina-specific expression of BMP-4 has been implicated in the dorsalization in the retina (5). However, no BMP-neutralizing factor has been identified in the retina so far: Neither noggin nor chordin was detected in the retina at E3 (stage 16 to 17) by in situ hybridization (10). The complementary expression patterns of Ventroptin andBMP-4 [Fig. 2, B and C, and Web fig. 2 (8)] suggest that Ventroptin prevents BMP-4 from affecting the ventral part of the retina to ensure the ventral cell fate. Because these are secretory molecules, they are expected to diffuse for a distance and interact with each other. We thus tested the effects of misexpression of Ventroptin and BMP-4 on each other's expression in the retina by in ovo electroporation [details are described in (8)]. This method allows us to express the transgenes almost uniformly in the retina [Web fig. 3 (8)]. When Miw/VOPT (8) was electroporated into the optic vesicle at stage 8 to 10, BMP-4 expression was markedly reduced at E3 (stage 18 to 20) (Fig. 3A, a; 5 out of 7 embryos). On the other hand, misexpression of BMP-4 repressed expression ofVentroptin (Fig. 3A, b; 7 out of 8 embryos). This is consistent with the findings that BMP-4 and its antagonists repress each other in early Xenopus embryos (15,16). These results suggest that Ventroptin and BMP-4 interact in vivo to keep their countergradient expression pattern along the D-V axis. There is a similar counteraction between Tbx5 and cVax (4, 5).

Figure 3

(A) Counteraction between Ventroptin and BMP-4. (a) BMP-4 expression at E3 (stage 18) in the eye electroporated with Miw/VOPT. (b)Ventroptin expression at E3 (stage 20) in the eye electroporated with RCAS/BMP-4. Insets in (a) and (b) show the expression of BMP-4 and Ventroptin, respectively, in the control eyes. Arrowheads in (a) and (b) indicate the dorsal and ventral retina, respectively. (B) Effects ofVentroptin misexpression on expression ofTbx5 and cVax. (a) Expression pattern of Tbx5 in the control eye. (b) Tbx5expression at E3 (stage 19) in the eye electroporated with Miw/VOPT. (c) Expression pattern of cVax in the control eye. (d) cVax expression at E3 (stage 19) in the manipulated eye. The positive domain expanded to the dorsal retina. Arrowheads in (a) through (d) indicate the dorsal retina.

Because BMP-4 is known to up-regulate Tbx5 and down-regulatecVax (5), we tested the effects of misexpression of Ventroptin on Tbx5 and cVaxexpression in the retina. After misexpression of Ventroptin,Tbx5 expression was reduced (Fig. 3B, a and b; 11 out of 14 embryos) and cVax expression was induced (Fig. 3B, c and d; 8 out of 11 embryos) in the dorsal retina at E3 (stage 18 to 20). The extent of these phenotypes was correlated with the dose of the electroporated DNA (10).

Misexpression of Tbx5 or cVax in the retina alters the pattern of projection of the retinal axons to the tectum along the D-V axis (4, 5). Therefore, we analyzed retinotectal projections at E18 to E19 after misexpression ofVentroptin. To observe the behavior of the retinal ganglion cell axons, we labeled a small number of the dorsal retinal fibers with 1,1′-dioctadecyl-3,3,3′,3′-tetramethyl-indocarbocyanine perchlorate (DiI) (Fig. 4A, c) (3). In the control animals, the dorsal retinal axons converge at a terminal zone in the middle of the ventral tectum (asterisk) on the contralateral side (Fig. 4A, b and e; 12 out of 12 embryos). In embryos with misexpressed Ventroptin (Fig. 4A, a and d), in contrast, the dorsal axons did not form a tight terminal zone at the proper position (asterisk), and their trajectories shifted to the dorsal side in the ventral tectum (small arrows). Moreover, numerous axons projected to the dorsal tectum (large arrows) (9 out of 11 embryos): Three of nine embryos showed projections of dorsal axons only to the dorsal tectum. The axons did not stop in the middle of the tectum but extended to the posterior end (arrowheads in Fig. 4A, a). These results differ from those of ectopic expression ofcVax in the dorsal retina, in which dorsal axons occasionally invade the dorsal tectum but never project beyond the middle of the tectum (4). To confirm mistargeting along the A-P axis, we also labeled the dorsotemporal axons. All these axons again overshot to the posterior end of the dorsal tectum in a manner similar to the dorsal axons (5 out of 5 embryos) (10,17). Therefore, aberrant projections in the embryos with misexpressed Ventroptin do not simply reflect the effect of dorsal induction of cVax.

Figure 4

(A) Retinotectal projection at E18 to 19 after simultaneous electroporation of Miw/VOPT and RCAS/VOPT. (a) A typical projection pattern in the embryo with misexpressed Ventroptin. Many dorsal axons shifted to the dorsal side (small arrows) or extended on the dorsal tectum (large arrows), and all the axons projected to the posterior end of the tectum (arrowheads). The asterisk indicates the proper terminal zone in the control embryo. (b) A typical projection pattern in the control embryo. The axons formed a tight terminal zone (asterisk) in the tectum. (c) The arrow indicates the position of the DiI label in the dorsal periphery of the right retina, close to the A-P midline. (d and e) Schematic drawings of (a) and (b), respectively. Anterior is down. (B) Effects of misexpression of Ventroptin and noggin on the expression of ephrin A2. (a) Control eye.Ephrin A2 is not expressed in the temporal retina. (b) Eye transfected with Miw/VOPT and RCAS/VOPT.Ephrin A2 expression was induced mainly in the ganglion cell layer in the E8 temporal retina. (c) Eye transfected with RCAS/Noggin. Ectopic ephrin A2 expression was induced similarly. The right small panels in (a) through (c) are enlargements of the temporal areas boxed in the left panels. Scale bars, 600 μm. Nasal and temporal are top and bottom, respectively.

The temporal high–nasal low gradient expression of EphA3 receptor in the retina is responsible for the topographic retinotectal projection along the A-P axis (18, 19); and the nasal high–temporal low gradient of its ligands, ephrins A2 and A5 in the retina, is involved in this process by modulating EphA receptor function (20, 21). We examined the effect ofVentroptin misexpression on the expression patterns of these genes. Overexpression of Ventroptin induced expression ofephrin A2 [which was not expressed in the temporal retina of the control eye (Fig. 4B, a)] in the temporal retina mainly in ganglion cells (Fig. 4B, b; 6 out of 6 embryos); whereas we did not detect any obvious alteration in the expression patterns ofEphA3 and ephrin A5 (10). The ectopic projection of the dorsal and dorsotemporal axons to the caudal end of the tectum is explained by this ephrin A2 induction:Ephrin A2 overexpression in the retina possibly modified the signal transduction capacity of EphA receptors to make them insensitive to ephrins in the posterior tectum (20). CBF-1,CBF-2, SOHo1, and GH6 are known to be involved in the retinal specification along the A-P axis and show asymmetric distributions along the A-P axis in the retina far earlier than Ventroptin (3, 6).Ventroptin misexpression did not alter the expression patterns of these transcription factors (10). On the other hand, SOHo1 and GH6 do not affect the expression of ephrin A2 (6), which suggests thatVentroptin is not controlled by these two factors.

The polarity along the D-V axis in the retina appears to be determined after stage 11 and before stage 13/14 in the chick (22,23). BMP-4 and Ventroptin expressions are detectable in the optic vesicle from stage 10 or 11 onward (5, 10). Therefore, the counteraction between Ventroptin and BMP-4 appears to determine and maintain the regional specificity along the D-V axis. At E6, when the first retinal axons enter the tectum (24), Ventroptin shows the nasal high–temporal low gradient expression pattern. From this stage on, Ventroptin seems to control retinotectal projection along the A-P axis by controlling the expression of ephrin A2.BMP-4 is expressed specifically in the dorsal retina, evenly along the A-P axis. At later stages (E6 to 8), expression ofBMP-4 was markedly reduced and was detected only in the peripheral margin of the retina (10). Thus, BMP-4 is not likely to be involved in the projection along the A-P axis. On the other hand, we found that Noggin, a structurally unrelated BMP antagonist, had the same activity as Ventroptin in expression ofTbx5 (10), cVax (10), andephrin A2 (Fig. 4B, c; 6 out of 6 embryos), when it was misexpressed in the retina. These results suggest the presence of another member of the TGF-β family in the retina, which binds to Ventroptin (and Noggin) and is involved in retinotectal projection along the A-P axis. Our study thus indicates that BMP family members and Ventroptin are involved in topographic retinotectal projection along the D-V and A-P axes.

  • * To whom correspondence should be addressed. E-mail: madon{at}


Stay Connected to Science

Navigate This Article