Tbx5 and the Retinotectum Projection

See allHide authors and affiliations

Science  07 Jan 2000:
Vol. 287, Issue 5450, pp. 134-137
DOI: 10.1126/science.287.5450.134


Dorsal and ventral aspects of the eye are distinct from the early stages of development. The developing eye cup grows dorsally, and the choroidal fissure is formed on its ventral side. Retinal axons from the dorsal and ventral retina project to the ventral and dorsal tectum, respectively. Misexpression of the Tbx5 gene induced dorsalization of the ventral side of the eye and altered projections of retinal ganglion cell axons. Thus, Tbx5 is involved in eye morphogenesis and is a topographic determinant of the visual projections between retina and tectum.

Dorsal (medial) and ventral (lateral) aspects of the eye are distinct from early stages of development, and retinal axons project in an organized topographic manner (1, 2). ChickTbx5 gene, a member of the T-box transcription factor family, is expressed in the dorsal side of the developing eye (3) and is homologous to the Drosophila optomotor blind (omb) gene, which regulates optic lobe development (4). Therefore, the Tbx5 gene could be a good candidate for a determinant of the dorsal-ventral (D-V) axis of the eye.

Tbx5 expression in the chick eye was first detected at stage 11 throughout the retina, with the strongest signal in the dorsal retina (Fig. 1A). Expression was later confined to the dorsal eye cup (Fig. 1, B and C) (5).Tbx5 is expressed in both retinal pigment epithelium (RPE) and neural retina (NR) at these early stages. Although expression in RPE started to fade out at embryonic day (E) 10, robust expression was maintained in the dorsal NR (all layers except the nerve fiber layer at the inner surface of the retina; Fig. 1, D to F).

Figure 1

Expression patterns of Tbx5 gene in the developing eye. (A) Expression of Tbx5 gene is detected in a broad area at stage 11. (B) At stage 14, expression becomes restricted to the dorsal half. (C)Tbx5 is expressed in a graded fashion with the strongest signal at the dorsal-most end. (D to F) At E10, dorsal expression is maintained only in the NR. Tbx5transcripts were never detected in the ventral retina except for a few cells at the inner surface of the NR (F). (G andH) Misexpression of the Tbx5 gene in the ventral side induces dorsalization of the ventral eye. This eye extends both dorsally and ventrally (H). The lens was shifted dorsally in this oval-shaped eye, whereas the normal eye is round and the lens is located ventrally (G). Red lines indicate the midline of the oval-shaped eye (H). Bars in (G) and (H), 1 mm.

To analyze the function of Tbx5 in D-V axis determination, we used in ovo electroporation to express the Tbx5 gene in the ventral eye (6, 7). This method allows temporary site-directed expression of foreign genes. We used plasmid CAGGS-Tbx5-EGFP, which expresses Tbx5 under the control of the cytomegalovirus-β actin hybrid promoter (8). Expression of the transgene can be monitored by green fluorescent protein (GFP) fluorescence derived from Tbx5-EGFP fusion protein. When this plasmid was electroporated to the eye vesicles of stage 8 chick embryos, oval-shaped eyes were formed, presumably because of elongation of the eye cup in both dorsal and ventral directions (Fig. 1H). In these eyes, the location of the lens was shifted dorsally.

Pax2 is involved in the formation of the ventral structures (9). When CAGGS-Tbx5-EGFP was electroporated into the ventral half (Fig. 2D), expression ofPax2 was repressed (Fig. 2E). Chicken Vax gene expressed in the ventral side of the developing eye cup (10) was also repressed (Fig. 2B). In addition, ventrally expressedEphB2 and EphB3 were also repressed (11). In contrast, the EphrinB1 andEphrinB2 genes were expressed only in the dorsal side (Fig. 2, I and K). When Tbx5 was misexpressed, bothEphrinB1 and EphrinB2 genes were induced in the ventral retina (Fig. 2, H and J). Thus, misexpression ofTbx5 in the ventral side of the eye results in repression of the ventral markers Pax2, Vax, EphB2, and EphB3 and induction of the dorsal markersEphrinB1 and EphrinB2 with dorsalization of the ventral side of the eye.

Figure 2

Misexpression of Tbx5 gene represses ventral markers Pax2 and Vax and induces dorsal markers EphrinB1 and EphrinB2. Misexpression was monitored by GFP signals derived from the Tbx5-EGFP fusion gene (A) and the section hybridization (D andG). (B) Vax expression was repressed by misexpression of Tbx5 (an arrowhead), whereas the control eye shows the robust expression in the ventral half (C). (E) Pax2 was also repressed, whereas the control eye shows the normal expression (F). (H andJ) Tbx5 induces dorsal markers EphrinB1 andEphrinB2, which are not expressed in the ventral side of the normal eye cup (I and K).

Between E6 and E12, axons of retinal ganglion cells begin to enter the optic tectum and establish synaptic connections (2). To analyze possible involvement of Tbx5 expression on the formation of retinotectum projections, we injected RCASBP-Tbx5-EGFP retrovirus into the optic cups at stage 8 to 10 (12) and then analyzed both retinas and tectums by anterograde DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) labeling (13, 14). Virus injection was chosen because rapid transgene expression by electroporation induces dorsalization of the ventral eye, which might interfere with the coordinated exit of retinal axons into the optic stalk. Injection of retrovirus did not induce morphological changes, perhaps because there is a time lag (around 16 hours) between injection of virus and the full expression of genes (15).

Ventral retinal axons normally converge on a terminal zone in the retinorecipient layer of the dorsal side of the tectum (n = 15) (2) (Fig. 3A). In contrast, in 7 out of 30 embryos infected with RCASBP-Tbx5-EGFP virus, axons did not make a tight focus (Fig. 3B), converging on two aberrant spots where abnormal arborization and synaptic varicosities were observed. Some axonal trajectories were found to be deflected from their normal path (11). From these results, we conclude that misexpression of the Tbx5gene in the ventral side of the retina changes the pattern of axonal projection in the tectum.

Figure 3

Anterograde and retrograde DiI labeling. (A) Normal ventral retinal axons enter the dorsal surface of the tectum and make a tight focus on the retinorecipient layer, indicated by an asterisk. (B) When Tbx5 was misexpressed, ventral retinal axons did not converge on an appropriate spot (indicated by an asterisk). Instead, they passed it and converged to two aberrant spots with abnormal arborization and synaptic varicosities. Some axons were found in the opposite dorsal side of the tectum (red arrowhead). (C and D) Schematic drawings of (A) and (B), respectively. Caudal to the left, rostral to the right. (E to J) Retrograde DiI labeling. In normal retina, DiI signals were found exclusively in the dorsal half (E) and not in the ventral half (H). When Tbx5-EGFP was misexpressed in ventral retina, GFP signals were observed in the ventral retina (J) and not in the dorsal retina (G). In such retina, DiI signals were found in both dorsal (F) and ventral retina (I). (K) Schematic representation of the areas analyzed (small rectangles). N, nasal; T, temporal. (L) A spot made by DiI injection is shown (indicated by a red arrow). cr, cerebrum; cl, cerebellum; e, eye; t, tectum.

To investigate this effect further, we also studied axonal projection by retrograde DiI labeling (2). DiI dissolved inN,N-dimethylformamide was injected once with a sharp glass pipette into the ventral tectum to label a small subset of retinal axons (Fig. 3L). After injection, the embryos were incubated for 2 days to allow DiI to be carried back to the retinal ganglion cells, and then dissected retinas were analyzed (Fig. 3K). In control embryos (n = 10), 226 (98%) labeled retinal cells were found in the dorsal retina and 4 (2%) labeled cells were found in the ventral retina (Fig. 3, E and H). After injection of Tbx5-EGFP virus, there was a marked increase of DiI-labeled cells in the ventral retina (31%) (4 out of 10 embryos) (Fig. 3, F and I); 223 DiI signals were found in the dorsal retina, whereas 101 were found in the ventral. Expression of Tbx5 was confirmed by GFP fluorescence (Fig. 3J).

Even with ectopically expressed Tbx5, some ventral retinal axons projected normally to the dorsal side of the tectum. This may have been because ectopic expression of the Tbx5 gene occurred too late or not all of the retinal cells were infected. In fact, the amount of Tbx5 retrovirus injected was limited, because excessive expression of Tbx5 caused deformities of the brain and premature death of embryos (16). Consequently, injected retina did not show ubiquitous expression of Tbx5-EGFP.

Our experiments demonstrate that misexpression of the Tbx5gene can induce dorsalization of the eye, suggesting that Tbx5 is involved in the early determination of the D-V axis of developing eye, possibly in conjunction with the dorsally expressed genesEphrinB1 and EphrinB2 and the ventrally expressed genes EphB2, EphB3, Pax2(9), and Vax (10). Our data suggest that both Pax2 and Vax act downstream of Tbx5, as both genes were repressed by misexpression of Tbx5. However, no effect was observed on the expression of Pax2 and Vax in the optic stalk (17). Pax2 may therefore not be a direct target but may be controlled by other factors in the developing eyes. One candidate could be retinoic acid (RA) because implantation of RA-soaked beads induces ventralization and dorsalpax[b] expression in zebrafish eyes (18). Other factors may also be involved in the determination of the D-V axis of the eye because dorsalization and repression of Pax2 andVax were not complete. Such factors could be bone morphogenetic proteins (BMPs), which are also involved in the D-V axis determination (19). Moreover, Drosophila omb acts in conjunction with decapentaplegic (dpp), which belongs to the same family of BMP molecules.

To investigate the functional roles of BMPs, we electroporated CAGGS–mouse BMP4 in the ventral side of eye cups. When mouse BMP4 was misexpressed in the ventral half, round eyes were formed with expansion of Tbx5 expression in the ventral half (Fig. 4, A and B). In such eyes, expression of Vax and Pax2 was repressed (Fig. 4, D and F). As shown in Fig. 4, B, D, and F, misexpression of BMP4 induced more profound effects on eye morphology. These observations indicate that BMP4 acts upstream of Tbx5 (Fig. 4G).

Figure 4

When mouse BMP4 was misexpressed in ventral sides of eye cups, Tbx5 was induced [red arrowheads in (B)] and both Vax and Pax2 were repressed [red arrowheads in (D) and (F), respectively]. Normal expression patterns of these genes are shown in (A), (C), and (E). Misexpression of BMP4 induces round eyes and deformities of ventral structures [(B), (D), and (F)]. The signaling cascade is illustrated in (G). BMP4 in the dorsal side of the eye cup induces Tbx5. Tbx5 inducesEphrinB1 and EphrinB2 and represses thePax2 and Vax genes. This inhibitory effect establishes the mutually exclusive patterns of the dorsal and ventral markers.

Our results have shown that the Pax2, Vax,EphrinB1, and EphrinB2 genes act downstream of the Tbx5 gene. This genetic hierarchy seems in good accordance with the serious expression of BMP4,Tbx5, Pax2, EphrinB1, andEphrinB2 (stage 10+, 11, 11, 13 to 15, and 13 to 15, respectively) (2, 3, 5,10). Misexpression of Tbx5 in the ventral side of the eye induces marked changes of the retinotectum projection without obvious morphological alteration. This rules out the possibility that the effect of Tbx5 misexpression on the projection is secondary to the dorsalization, because the exit of retinal axons into the optic nerve and the optic nerves themselves were formed normally in virus-infected eyes. Hence, we conclude that the signaling cascade mediated by Tbx5 plays a key role in both eye morphogenesis and the visual projection.

  • * These authors contributed equally to this work.

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


Stay Connected to Science

Navigate This Article