Expression of Drosophila omb-Related T-Box Genes in the Developing Human and Mouse Neural Retina

Abstract

purpose. To examine the role of Drosophila optomotor blind (omb)–related T-box genes in development of human and mouse retina.

methods. Mouse Tbx2, Tbx3, and Tbx5 and human TBX2 cDNAs were isolated from retinal cDNA libraries by hybridization to the Drosophila omb gene. Gene expression patterns in developing retina were analyzed by in situ hybridization.

results. TBX2/Tbx2, TBX3/Tbx3, and TBX5/Tbx5 were expressed asymmetrically across the embryonic neural retina with highest levels of mRNA within dorsal and peripheral retina. The dorsoventral gradient of TBX2 expression disappeared before the ganglion cell layer (GCL) formed. Its expression then became restricted to the inner neuroblastic retina and later to the GCL and inner nuclear layer (INL). The dorsal expression domains of TBX5/Tbx5 and TBX3/Tbx3 were maintained during formation of the GCL. As the retina matured, TBX3/Tbx3 expression was restricted to the INL, and TBX5/Tbx5 was expressed within the GCL.

conclusions. The expression pattern of TBX2, TBX3, and TBX5 within the developing retina supports the idea that the encoded transcription factors play a role in providing positional information important for topographic mapping and in differentiation of distinct cell types across the laminar axis of the retina.

The mature visual system comprises a complex network of neuronal connections whose physiology and axonal projections are well characterized. In the human retina, more than 100 million rod and cone photoreceptors transmit visual information to some 1.2 million ganglion cells through the bipolar, amacrine, and horizontal interneurons of the inner nuclear layer (INL). ¹ The ganglion cell axons project with precise topographic mapping to synaptic targets within the higher visual system. Recent progress has being made in identifying the guidance molecules that ensure the ganglion cell axons navigate to appropriate synaptic targets during development. ^{2 3 4} Less is known, however, about the genetic mechanisms by which ganglion cells and the retinal interneurons achieve their differentiated identities and make precise synaptic connections during development. 

Although the visual systems of the cat and the chick have been extensively characterized, these systems are not amenable to genetic analysis. In contrast, the knowledge of the genetics of the visual system of the fruit fly, Drosophila melanogaster, is relatively advanced because of the ease of generating mutant flies. ⁵ An emerging strategy that is useful for the identification of genes regulating development of the mammalian visual system is to analyze homologues of Drosophila genes that are important in visual system development in the fly. ⁶ This strategy is based on accumulating evidence that demonstrates the conservation of genetic mechanisms underlying homologous structures in diverse species and the discovery that mutations in conserved genes underlie inherited eye malformations in flies and mammals. ⁷ For example, there is evolutionary conservation of the transcription factors, Pax6 and Sine oculis, which are important in eye development, and of opsin proteins involved in phototransduction. ⁸  

The pathophysiology resulting from mutation of the Drosophila gene optomotor blind (omb) demonstrates that the omb gene is an important regulator of neuronal cells, which process and integrate visual information from the compound eye and transmit it to the central brain. ⁹ In the third larval instar of the fruit fly, omb is expressed in the neuroblasts of the optic lobe, which differentiate into ganglionic neurons with the arrival of axons projecting from the photoreceptor cells. ¹⁰ Several omb alleles are pupal lethal, and optic lobe neurons fail to differentiate. Other alleles give rise to viable flies with reduced levels of omb protein, resulting in specific behavioral and neuroanatomic defects in the visual system. ⁹ These mutant flies do not have large fibers in the lobula plate ganglionic complex of their visual system. 

OMB is of particular interest, because it is a member of an important family of developmental regulators encoded by T-box genes. ^{11 12} These genes encode transcription factors that share a highly conserved and novel type of DNA-binding domain, the T-box domain. T-box genes have been characterized in a range of species, including humans, indicating the evolutionary conservation of this gene family. Phylogenetic analysis predicts that most animal species have at least five T-box genes. ¹² Mutations in several T-box genes cause a variety of developmental defects in mice and humans, thus highlighting the vital role that T-box genes play in diverse developmental processes, including limb and heart morphogenesis and mesoderm formation. ^{13 14 15 16 17 18}  

We have focused on the analysis of omb-related human and mouse genes to assess whether these genes play significant roles in development of the mammalian retina. Human and mouse retinal cDNA libraries were screened using the T-box region of the omb gene as a hybridization probe to identify omb-related genes expressed in the retina. We examined expression of the three T-box genes encoding the transcription factors TBX2, TBX3, and TBX5 and found that each gene is expressed in an overlapping domain within the dorsal embryonic neural retina. As the retina differentiates, each gene is expressed in restricted and distinct subsets of retinal cells. 

Methods

In Situ Hybridization

All animal procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Mouse embryos were obtained from matings of C57BL/6 x CBA mice. The day on which the vaginal plug was detected was designated embryonic day (E)0.5. Eyes from embryos at E10.5, E12.5, and E14.5 and from adult mice were analyzed. After informed consent and ethical permission had been sought and granted, human embryonic and fetal eye specimens were obtained from the Medical Research Council Tissue Bank and the Human Developmental Biology Resource, United Kingdom, according to the Polkinghorne Guidelines of the United Kingdom, which are in line with the tenets of the Declaration of Helsinki. Embryonic and fetal ages in weeks after conception were determined, either from hand and foot measurements, or, for older fetuses, by subtracting 2 weeks from the time since the last menstrual period. Tissues were fixed overnight in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C. Standard procedures were used to embed specimens in paraffin wax and for hybridization of ³⁵S-radioisotope–labeled riboprobes to tissue sections. ¹⁹ For human specimens, optimal hybridization signal was achieved, using cryosectioned tissue and nonradioactive digoxygenin-labeled riboprobes. After fixation, human eye specimens were placed in 20% sucrose in PBS for 24 hours, oriented dorsoventrally in optimal cutting temperature (OCT) compound and then flash frozen using isopentane and dry ice. Cryosections (10μ m) were mounted onto 3-aminopropyltriethoxysilane (TESPA; Sigma, St. Louis, MO)–coated slides. In this study eyes at 6 weeks (n= 3), 8 to 9 weeks (n = 4), 12 to 13 weeks (n= 4), and 15 weeks (n = 2) were analyzed. Cryosections were hybridized with 1 ng/μl digoxygenin-labeled riboprobe, in hybridization buffer (1 mg/ml transfer [t]RNA, 50% formamide, 10% dextran sulfate, 1× Denhardt solution, 200 mM NaCl, 5 mM NaH₂PO₄, 5 mM Na₂HPO₄, 5 mM EDTA, and 10 mM Tris [pH 7.5]) for 16 hours at 65°C. After hybridization, sections were washed at 65°C, in 1× SSC, 50% formamide, and 0.1% Tween-20 two times for 1 hour each, then twice for 30 minutes in maleic buffer with Tween; 0.1 M maleic acid, 0.15 M NaCl, 0.1% Tween 20 (MABT; Roche Molecular Biochemicals, Indianapolis, IN) at room temperature. To visualize the hybridized probe, sections were blocked with 1× MABT-2% blocking reagent (Roche) and 20% heat-inactivated sheep serum for 1 hour and incubated with anti digoxygenin alkaline phosphatase–conjugated antibody (1:1000 dilution) at 4°C overnight. Sections were washed four times for 5 minutes each in MABT, two times for 10 minutes each in staining buffer (2% 5 M NaCl, 5% 1 M MgCl, 10% 1 M Tris [pH 9.5], 0.1% Tween-20), and the color reaction was performed with 340 μg/ml nitroblue tetrazolium (NBT) and 170 μg/ml 5-bromo-4-chloro-3-indoyl phosphate (BCIP) in staining buffer in the dark. Sections were then mounted (Vectamount; Vector, Burlingame, CA) for microscopy. 

The procedure used for wholemount in situ hybridization was as previously described, ¹⁹ except a modified hybridization buffer was used: 50% formamide, 5× SSC, 2% blocking powder, 0.1% Triton X-100, 0.5% 3-([3-cholamidopropyl] dimethylammonio)-2-hydroxy-1-propanesulfonate (CHAPS), 1 mg/ml yeast RNA, 5 mM EDTA, 50 μg/ml heparin. Hybridizations were performed at 80°C and posthybridization washes were as follows: 5 minutes in solution 1 (50% formamide, 5× SSC, 0.1% Triton X-100, 0.5% CHAPS); 5 minutes in three parts solution 1 to one part 2× SSC; 5 minutes in one part solution 1 to one part 2× SSC; 5 minutes in 1 part solution 1 to three parts 2× SSC; two times for 30 minutes each in 0.1% CHAPS and 2× SSC; and two times for 30 minutes each in 0.1% CHAPS and 0.2× SSC. All washes were performed at 65°C. Four E10.5 embryos were hybridized with each antisense and sense mouse probe in wholemount experiments. 

The following plasmids were used to synthesize riboprobes for in situ experiments: (1) 498 bp of human TBX2 cDNA downstream of the T domain in pGEM3Zf(+) (Promega, Southampton, UK); (2) 1.4 kb human TBX3 cDNA in pBluescript II-SK(−) (Stratagene, La Jolla, CA); (3) 2.4 kb human TBX5 cDNA in pBluescript II-SK(−); (4) 255 bp of the T-box region of mouse Tbx2 cDNA in pBluescript II-KS(−); (5) 255 bp of the T-box region of the mouse Tbx3 cDNA in pBluescript II-KS(−); (6) 1.1 kb of Tbx5 cDNA in pBluescript II-KS(−). Sense and antisense RNA transcripts were synthesized, using either SP6 polymerase (Roche) or T7 polymerase (Roche), and labeled with ³⁵S-uridine triphosphate (UTP) or digoxygenin-UTP for hybridization to 8-μm tissue sections or wholemount embryos. To compare expression patterns of TBX2/Tbx2, TBX3/Tbx3, and TBX5/Tbx5, serial sections from human and mouse eye specimens were hybridized with respective probes. No signal was detected with the RNA sense probes. Embryos and sections were photographed with a photomicroscope (Diaplan; Leica, Cambridge, UK) or a stereomicroscope (MZ12; Leica), respectively (Ektachrome 64T film; Eastman Kodak, Rochester, NY), and the images were digitized on a scanner (FilmScan 200; Epson Seiko, Nagano, Japan) and assembled into figures on computer (Photoshop ver. 5.0; Adobe, San Diego, CA, and Powerpoint; Microsoft Corp. Redmond, WA). 

Isolation of Mouse T-Box cDNAs

A ³²P-labeled 1.8-kb BamHI fragment from the Drosophila omb cDNA, including the 600-bp T-box domain was used as a hybridization probe to screen a cDNA library in λZAP Express vector (Stratagene) prepared from neural retina and lens dissected from E15.5 mouse embryos; 5 × 10⁵ plaque-forming units were hybridized overnight at reduced stringency at 52°C in a standard hybridization solution. Filters were washed twice in 0.5× SSC and 0.5% SDS at 52°C for 20 minutes before exposure to x-ray film at −80°C. pBK-CMV plasmids were excised from positive hybridizing plaques identified in tertiary screens and sequenced directly using plasmid primers flanking the insertion site and a dye terminator kit (Big Dye; Applied Biosystems, Foster City, CA). Of 14 clones analyzed, three contained T-box cDNAs. The other 10 clones that hybridized to the 1.8-kb omb probe did not hybridize to a smaller 300-bp omb probe, covering only the central T-box sequence, and encoded cDNAs for unrelated proteins, mouse lens α-crystallin, a protein of the major histocompatibility class II complex, and a protein-elongation factor. 

Degenerate PCR amplifications were performed using 1 μg mouse E15.5 retina and lens cDNA library in plasmid pBK-CMV. The degenerate primers span nucleotides encoding two conserved amino acid sequences, YIHPDSP and AVTAYQN, in the T-box of Drosophila omb and other T-box genes. PCR fragments of approximately 255 bp obtained after two 30-cycle rounds of amplification were gel purified using a gel extraction kit (QiaQuick; Qiagen, Crawley, UK) and subcloned into the pGEM-T vector (Promega) according to the manufacturer’s instructions. Recombinant plasmids were sequenced directly. 

Results

Identification of T-Box Genes Expressed in the E15.5 Mouse Retina

To isolate T-box genes, which are expressed in the developing retina and are related by sequence to the Drosophila omb gene, we used both conventional cDNA library screening and degenerate PCR amplifications. At E15.5 in the mouse, all six types of neuronal cell (rod, cone, horizontal, amacrine, bipolar, and ganglion) and Müller glial cells, which are characteristic of the mature neural retina, are developing and differentiating. ¹ A³²P-labeled cDNA probe that spans the T-box domain of the omb gene was used to screen a cDNA library prepared from E15.5 mouse neural retina and lens at low stringency. Four positive clones were identified and sequenced. Sequence comparisons indicated that three clones contained sequences identical with the Tbx2 cDNA previously isolated from E11.5 mouse embryos, ¹¹ and one contained sequence identical with Tbx5 cDNAs isolated from E8.5 mouse embryos and embryonic mouse limb cDNA libraries (GenBank Accession No. U57330, AF140427 ^{20 21} ). The 5′ ends of the retinal Tbx2 cDNAs lie within a trinucleotide repeat sequence (CGG)₅ that encodes a string of alanine residues (starting at position 228 bp of the previously reported mouse Tbx2 cDNA sequence, U15566. The retinal Tbx5 cDNA contains a 129-bp 5′ untranslated leader sequence (compared with 418 bp in the limb Tbx5 cDNA) and a 1005-bp 3′ untranslated region. 

Conventional library screening was complemented by performing PCR amplifications of aliquots of the same E15.5 eye cDNA library using degenerate primers that encode two conserved hexapeptide stretches within the T-box region of Drosophila omb (M81796) and other T-box genes. PCR products of approximately 255 bp were subcloned, and 33 recombinants were sequenced. Of these, four were identical with mouse Tbx3 cDNA (U57328) previously isolated from E11.5 embryos, ¹¹ and 29 were identical with mouse Tbx2. 

We concluded that three omb-related T-box genes, Tbx2, Tbx3, and Tbx5, are expressed in the developing eye (neural retina and lens). No other T-box cDNAs were detected. Across the 180-amino-acid (aa) region of the omb T-box domain (nucleotides [nt] 868-1408; Y16899), the mouse Tbx2, Tbx3, and Tbx5 cDNAs share 66.9% nt (72.8% aa), 64.0% nt (72.6% aa) and 64.6% nt (70.2% aa) identity, respectively. Of these, the Tbx2 gene is most highly related at the amino acid and nucleotide level to the Drosophila omb gene. 

Identification of Human Homologues of omb

Expressed sequence tag (EST) databases were also searched for human cDNAs sharing high levels of sequence identity with the Drosophila omb cDNA. A human cDNA clone (IMAGE ID:223216; available from IMAGE Consortium at http://[email protected]) which shares sequence similarity with omb, was identified from the Soares adult retinal cDNA library (Washington University School of Medicine; N2b5HR, prepared from a 55-year-old male retina). The clone contains a 2.6-kb insert, and sequence analysis confirmed that it represents the human TBX2 mRNA (NM005994). The 5′ end of the human retinal TBX2 cDNA, like the mouse Tbx2 retinal cDNAs, lies within a conserved trinucleotide repeat encoding alanine residues. Because these human and mouse retinal TBX2/Tbx2 cDNAs terminate at their 5′ ends within a guanosine cytosine (GC)-rich segment interrupting the T-box domain and are shorter than cDNAs isolated from other sources, it is likely that they do not represent full-length mRNAs. No other human T-box cDNA sequences derived from retinal RNA were identified within the EST databases. 

In Situ Hybridization Analysis of TBX2, TBX3, and TBX5 Expression in the Neural Retina

To assess whether T-box genes are involved in the differentiation of retinal neurons, we used in situ hybridization techniques to examine gene expression during development of the mammalian retina. This study focused on the analysis of human TBX2 and its mouse homologue Tbx2 during retinal development, because the cDNA library screening experiments identified TBX2/Tbx2 cDNA clones within the developing and mature retina. Human cDNA probes for TBX3 and TBX5, the human homologues of mouse Tbx3 and Tbx5 (kindly provided by David Law, University of Michigan, Ann Arbor, MI and J. David Brook, Queen’s Medical Centre, University of Nottingham, UK, respectively) were obtained for a comparative expression study. Expression of TBX2/Tbx2 was compared with expression of TBX3 and TBX5 in human embryonic and fetal retina and with Tbx3 and Tbx5 in the differentiating and adult mouse retina. No previous studies have been performed to analyze expression of these genes within the human eye. Expression of Tbx2, Tbx3, and Tbx5 has been reported previously within the mouse optic cup, ²² but expression of these genes has not been analyzed during retinal differentiation and lamination. 

Expression of TBX2, TBX3, and TBX5 Expression across the Optic Cup

To examine patterns of gene expression across the developing optic cup, human embryonic and fetal eyes were sectioned either sagittally through the dorsal (superior) and ventral (inferior) retina (Figs. 1A 1E 1H) or coronally, revealing all four quadrants of the retina (nasal, temporal, dorsal, and ventral; data not shown). At 6 weeks after fertilization the embryonic neural retina was immature, consisting almost entirely of neuroblastic cells. Most of the characteristic cell types of the mature retina were as yet unborn; this period marks the beginning of retinal ganglion cell genesis. At this stage TBX2 was expressed in a dorsoventral gradient across the optic cup with the highest expression in the dorsal hemisphere (Fig. 1B) . Similar gradients of expression were seen for TBX3 and TBX5 (Fig. 1C 1D) . The TBX2 expression domain extended more widely across the dorsal hemisphere (Fig. 1B) and encompassed the smaller TBX5 domain, which is most abundant in the dorsal peripheral retina (Fig. 1D) . TBX2 expression subsequently lost dorsoventral asymmetry and was expressed throughout the inner neuroblastic (INB) retina from 8 to 9 weeks onward (Fig. 1F) , whereas both TBX3 and TBX5 expression continued to be restricted to the dorsal retina until at least 12 weeks (Figs. 1G 1I 1J 1K) . 

Expression of TBX2, TBX3, and TBX5 across the Laminar Axis of the Developing Neural Retina

By the end of the embryonic period of development (8 weeks) the INB and outer neuroblastic (ONB) layers could be clearly distinguished (Figs. 1E 1L) , TBX2 expression became restricted to the INB layer at this stage (Figs. 1F 1M) . TBX2 mRNA was also detected outside the retina within the developing cornea (Fig. 1F) . Neither TBX3 nor TBX5 showed any restriction across the laminar axis at 8 to 9 weeks, and both were expressed across the width of the dorsal peripheral retina (Fig. 1G and data not shown). 

By 12 to 13 weeks, the process of stratification of the neural retina was under way and the inner plexiform layer (IPL) had formed (Fig. 1N) . TBX2 mRNA was most abundant in the developing ganglion cell layer (GCL; Fig. 1O ). TBX3 mRNA, by contrast, was expressed exclusively in cells of the inner aspect of the dorsal neuroblastic retina (Fig. 1P , and low-magnification view, 1I). The TBX5 expression domain began to narrow across the laminar axis at this stage and became restricted to newly born ganglion cells in the dorsal retina (Fig. 1J) . 

At 15 weeks, the GCL was clearly established, and the presumptive INL was visible because of the increasing stratification of the outer retina (Fig. 1Q) . TBX2 mRNA was restricted to the GCL and to the developing INL, including cells at the inner aspect of the neuroblastic layer (Fig. 1R) . TBX3 expression was no longer restricted across the dorsoventral retinal axis and instead showed restriction across the laminar axis to cells of the nascent INL (Fig. 1S) . At this stage, TBX5 mRNA could not be detected within the neural retina. 

In the adult retina (Fig. 1T) , expression of TBX2 was maintained in the GCL and could be detected within cells of the INL (Fig. 1U) . No TBX2 expression was found in photoreceptor cells (Fig. 1U) . Neither TBX3 nor TBX5 was detectable within the adult retina by in situ hybridization. However, RNA PCR amplification of adult neural retina samples indicated that low levels of TBX3 and TBX5 mRNA were in fact present (Fig. 2) , suggesting that a small number of retinal cells may express these genes. 

Expression of Tbx2, Tbx3, and Tbx5 in the Developing Mouse Retina

In situ hybridization of wholemount E10.5 mouse embryos (equivalent to ∼4 weeks’ human gestation) was performed using digoxygenin-labeled Tbx2, Tbx3, and Tbx5 riboprobes. Tbx2 mRNA was detected in the dorsal hemisphere of the optic cup (Fig. 3A) , consistent with previous reports. ²² At this stage Tbx3 was also expressed within the dorsal portion of the optic cup, but at relatively low levels (Fig. 3B) . Tbx5 mRNA localized to a narrow band in the dorsal portion of the optic cup within the Tbx2 expression domain (Fig. 3C) . To compare these expression domains, adjacent coronal sections through the optic cup of E10.5 embryos were hybridized with the Tbx2, Tbx3, and Tbx5 ³⁵S-labeled riboprobes. The plane of section cut simultaneously through the dorsal (superior) and ventral (inferior) retina, which at this stage consisted almost entirely of undifferentiated progenitor cells. The expression domains of the three genes were coincident (Figs. 3D 3E 3F) . The Tbx5 domain was tightly demarcated and lay within the broader Tbx2 domain, which extended more ventrally, whereas Tbx3 expression was indistinguishable at this stage from that of Tbx2 (Figs. 3D 3E 3F) . 

To examine variation in T-box gene expression across the retinal axes as retinal ganglion cells develop, transverse sections through the eye at E12.5 and E14.5 were analyzed by in situ hybridization (Fig. 4) . By E12.5 in the mouse (equivalent to approximately 6 weeks of human development), cell differentiation in the neural retina was under way, and nerve fibers originating in the primitive ganglion cells were projecting toward the optic disc (Fig. 4A) . ²³ Tbx2 and Tbx3 mRNAs were abundant throughout the dorsal hemisphere (Figs. 4B 4C) , but could not be detected in ventral retina (data not shown). Tbx5 mRNA was restricted to the dorsalmost third of the optic cup and was particularly abundant peripherally (Fig. 4D) . The continued restriction of T-box gene expression to the dorsal retina at this stage of development and the tightly restricted domain of Tbx5 expression was consistent with the pattern of expression of the orthologous genes in the human retina. 

By E14.5 (approximately 7–9 weeks of human development) the external and internal neuroblastic layers of the neural retina could be distinguished histologically (Fig. 4I) . This time point is around the peak birth date of ganglion and amacrine cells. ¹ Tbx2 expression extended throughout the dorsal hemisphere. In transverse sections through the optic cup at the level of the optic nerve Tbx2 mRNA was concentrated within the INB retina (Fig. 4E) . The restriction to the INB retina was more marked dorsally (Fig. 4H) and was not apparent ventral to the optic nerve (Fig. 4K) . In contrast, Tbx3 mRNA levels were low within the central retina (Fig. 4F and data not shown). Strong Tbx3 signal was only detected within the peripheral retina (Fig. 4F) in a dorsotemporal location. Here, its expression overlapped with Tbx5 (Fig. 4G) . Expression of Tbx3 and Tbx5 in the peripheral edge or ciliary margin of the retina extended across the width of the retina (Figs. 4F 4G) . 

In addition to expression at the periphery, Tbx5, similar to Tbx2, was abundantly expressed at the location of the newly forming ganglion cells in the inner retina (Fig. 4G) . The Tbx5 expression pattern had a marked punctate appearance, suggesting only subsets of ganglion cells were labeled (Fig. 4G) . Tbx5 labeling was detected in developing ganglion cells throughout the dorsal retinal hemisphere (Figs. 4G 4J 4M) . The Tbx5 expression domain had thus extended ventrally, compared with earlier stages (Fig. 4D) . 

At E14.5, Tbx2 expression (Figs. 4E 4H 4K) , but not Tbx3 or Tbx5 expression, was also abundant in the developing cornea and throughout the margins of the optic cup. At the optic cup margins, Tbx2 expression was found both within the retina and in the overlying neural crest–derived mesenchyme that gives rise to the anterior segment of the mature eye. This observation is consistent with expression of TBX2 mRNA within the developing cornea during human ocular development (Fig. 1F) . Tbx2, but not Tbx3 or Tbx5 was also detected within the mesenchyme of the developing upper eyelids (Fig. 4L) . 

In the mature mouse retina (Fig. 4N) , as in the human retina, Tbx2 mRNA was most abundant within the GCL and INL, whereas the photoreceptor layer did not express Tbx2 at significant levels (Fig. 4O) . In comparison, Tbx5 mRNA was not detected in the mature retina. Tbx3 expression was detected within only a subset of cells of the INL (Fig. 4P) , the location of these cells was consistent with the sites of expression of human TBX3 at 15 weeks. 

Discussion

In the present study at early stages of human and mouse development, three T-box genes, TBX2/Tbx2, TBX3/Tbx3, and TBX5/Tbx5, that are closely related to the Drosophila omb gene, were expressed in overlapping domains within the dorsal neural retina of the embryonic optic cup (Table 1) . Recent reports of expression of the orthologous Tbx2, Tbx3, and Tbx5 genes within the dorsal optic cup of chick, ²⁴ Xenopus frog, ²⁵ and zebrafish embryos ^{26 27} confirm a high level of evolutionary conservation of these spatial patterns of gene expression across the dorsoventral axis of the vertebrate eye. Phylogenetic comparisons assign the vertebrate Tbx2, Tbx3, and Tbx5 genes to the same T-box gene subfamily as the invertebrate omb gene and suggest that these vertebrate genes arose by duplication of an ancient vertebrate omb-like gene sequence. ^{12 20} In the eye imaginal disc of Drosophila, the omb gene does not display asymmetric expression. ¹⁰ However, within leg and wing imaginal discs, omb expression is restricted to a dorsal compartment, ²⁸ suggesting some conservation of function for omb-related genes in patterning of invertebrates and vertebrates. Other embryologic studies have demonstrated that Tbx2, Tbx3, and Tbx5 play important roles in vertebrate limb morphogenesis, ^{29 30} and mutation of the human TBX3 and TBX5 genes both affect formation of the upper limbs. ^{13 14 15}  

Defining the dorsoventral and nasotemporal axes of the neural retina is an essential step in development of the visual system. The most critical requirement for positional information across the developing retina is to provide coordinate identities to retinal ganglion cells and activate appropriate guidance molecules, because their axons project to topographic targets within the superior colliculus (optic tectum) and lateral geniculate nucleus. ³¹ The early asymmetric expression patterns of the T-box genes reported here, TBX2, TBX3, and TBX5, and by others in lower vertebrates, implicate all three genes in dorsoventral patterning of the optic cup and in controlling ganglion cell axon guidance. Human retinal axons grow into the brain during the eighth week after fertilization, ³² and asymmetric expression of TBX3 and TBX5 is maintained beyond this stage; hence, the time frame of expression is appropriate for activating ganglion cell surface markers needed for axon guidance. Other types of retinal cells are also distributed asymmetrically across the retina, and their development is likely to respond to positional information across the retinal axes. For instance, the dorsal retinal hemisphere has relatively high numbers of middle-wavelength cone photoreceptor cells compared with the ventral retina. ³³  

Until recently, little was known about the genetic pathways that demarcate retinal territories and influence neuronal differentiation across the dorsoventral and nasotemporal axes of the neural retina. There are now a number of molecules that have been identified showing asymmetric patterns of expression across the nasotemporal axis and a smaller number showing asymmetric expression across the dorsoventral axis. The latter include Eph receptors and their ephrin ligands ^{34 35} and retinoic acid–synthesizing enzymes, ³⁶ in addition to transcription factors such as the T-box genes examined in this study. Only a handful of dorsally expressed transcription factors have been described previously. ^{37 38 39} Understanding the interactions between these different classes of molecules is key to understanding the genetic control of development of retinal connections. In this respect it is interesting to note that in vitro both Tbx2 and TBX3 function as transcriptional repressors, whereas other T-box proteins have transcriptional activator properties. ^{40 41 42}  

In the past year, two important studies have demonstrated the major role that asymmetrically expressed transcription factors play in activating the guidance molecules necessary for retinotopic projections. In chick embryos, misexpression of Tbx5 in the ventral retina leads to dorsalization of the ventral hemisphere and aberrant routing of the ventral projections, ⁴³ whereas misexpression of the emx-related homeobox gene Vax2 within the dorsal retina represses Tbx5 expression and causes targeting errors in dorsal projections. ⁴⁴ Ectopic Tbx5 expression represses expression of members of the family of receptor tyrosine kinases (Eph receptors), EphB2 and EphB3, and induces expression of the membrane-bound ligands (ephrins), ephrinB1 and ephrinB2. ⁴³ Other Eph molecules have already been shown to be critical for guidance and mapping of ganglion cell axons along the nasotemporal retinal axis. ^{2 3 4} It seems likely that the T-box proteins in combination with other transcription factors such as Vax2, may establish retinal projections along the dorsoventral axis through regulation of expression of Eph/ephrin molecules. 

As the retina matured, all three T-box genes examined showed distinctive spatial patterns of expression across the laminar axis of the retina. None of the genes was expressed within the photoreceptor cells; instead, expression was limited to cells of the forming INL and GCL. Expression patterns of the three genes in mouse and human were consistent at all stages examined, during the period when retinal cell differentiation, lamination, and ganglion cell axonogenesis occur. TBX5 was expressed within ganglion cells, whereas TBX3 labeled subsets of cells within the forming INL, and TBX2 was expressed within both the GCL and the INL. 

The similarity between mouse and human expression patterns supports the use of the mouse as a model system for future functional studies. A number of transcription factors have been identified that show restricted patterns of expression across the laminar retinal axis during mouse development. ^{45 46} Some of these factors have recently been shown to play critical roles in determining cell commitment and differentiation. ^{47 48 49} The T-box gene expression is complex, in that it showed graduated expression in two dimensions, both across the retina and across the retinal layers (summarized in Table 1 ). These findings suggest a dual role for omb-related T-box genes in the human and mouse developing neural retina, in early dorsoventral patterning, and in the process of lamination that accompanies the differentiation of retinal neuroblasts into the many cell types of the inner retina. 

 Supported by a Medical Research Council Career Development Award G120/191 (JCS). JKLH is supported by a Child Health Research Action Trust studentship, and MM was supported by Grant Me-1666 from the Deutsche Forschungsgemeinschaft.

 Submitted for publication December 5, 2000; revised June 22, 2001; accepted July 3, 2001.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked“ advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

TBX2, TBX3, and TBX5 expression in the human retina. In situ hybridization using digoxygenin-labeled riboprobes for TBX2, TBX3, and TBX5 as indicated (B–D, F, G, I–K, M, O, P, R, S, U). Sections adjacent to the experimental sections were stained with hematoxylin and eosin (A, E, H, L, N, Q, T). (A–J) Sections are oriented so the dorsal–superior retinal hemisphere is uppermost. Arrow: dorsal orientation. (A–D) Ocular sections, sagittal plane from human embryonic eyes at 6 weeks. TBX2, TBX3, and TBX5 were expressed in the dorsal and peripheral neural retina (B–D). (E–G) Nine-week fetal eye, sagittal plane. TBX2 was expressed in the INB retina (F), whereas TBX5 was restricted to the dorsal peripheral retina (G). (H–K) Twelve-week fetal eye, sagittal plane, plane of section in (H–J) is nasal to the optic nerve, and (K, dorsal is to the right) is through the optic nerve. TBX3 (I) and TBX5 (J, K) showed dorsally restricted expression within the neural retina. TBX5 expression was becoming restricted to the developing GCL (J). (L–U) High-power view of sections through retina showing TBX2 and TBX3 expression across the laminar axis. At 8 weeks, TBX2 expression was highest in the INB retina (M). At 12 to 13 weeks TBX2 was detected in the developing GCL (O), whereas TBX3 was restricted to the innermost cells of the neuroblastic layer (P and low-power magnification, I). At 15 weeks, TBX2 mRNA was detected in the GCL, and the developing INL (R), whereas TBX3 expression was restricted to the innermost cells of the developing INL (S). In the adult, TBX2 mRNA was detected in the INL and the GCL (U). Scale bar, (A–K) 200 μm; (L–S) 25 μm; (T, U) 10μ m.

Figure 2.

 

PCR amplification of TBX2, TBX5, and TBX5 mRNA from human retina. RNA PCR amplification of TBX2, TBX3, and TBX5 mRNA from 10-, 11-, and 14-week human fetal eyes and adult neural retina (NR). TBX2, TBX5, and TBX5 and ubiquitously expressed phosphoglucomutase, PGM1–specific primers, were used to amplify fragments of 209, 261, 295, and 417 bp, respectively. All primers are from exon sequences and PCR products span exon–intron boundaries. Thirty cycles of amplification were performed for each primer set to amplify aliquots of oligo dT-primed cDNA, under standard conditions. Lane M: molecular weight standard.

Figure 3.

 

T-box gene expression in the embryonic mouse optic cup. In situ hybridization of wholemount E10.5 embryos hybridized with Tbx2 (A), Tbx3 (B), and Tbx5 (C) digoxygenin-labeled riboprobes. Dotted line: plane of sections (D), (E), and (F). In situ hybridization of ocular sections from E10.5 embryos hybridized with Tbx2 (D), Tbx3 (E), and Tbx5 (F) using radioactively labeled riboprobes. Arrows: expression of Tbx2, Tbx3, and Tbx5 in dorsal hemisphere of the optic cup. Expression of Tbx2 and Tbx3 in the maxillary process (mx) and mandibular process (md) is also indicated. Scale bar, (A–C) 500 μm; (D–F) 200μ m.

Figure 4.

 

T-box gene expression in developing and adult mouse retinas. In situ hybridization analysis using ³⁵S-labeled riboprobes showing Tbx2, Tbx3, and Tbx5 expression in the E12.5 (B–D) and E14.5 (E–H, J–M) embryonic retina and adult mouse eye (O, P). (A), (I), and (N) are stained with hematoxylin and eosin. Arrow: nasotemporal axis. (A–D) Eye sections from E12.5 mouse, transverse plane. (A, B) Through the optic nerve; (C, D) through the dorsal retinal hemisphere. Tbx2 and Tbx3 were expressed in the dorsal retinal hemisphere (B, C). Tbx5 expression was restricted to the dorsalmost third of the optic cup (D). (E–M) Eye sections from E14.5 mouse, transverse plane; (E–G) through the optic nerve; (H–J) and (L–M) dorsal; (K) ventral to the optic nerve. Tbx2 was expressed in the INB of the dorsal retinal hemisphere (E, H) but not in the ventral hemisphere (K). Tbx2 was also expressed in the developing cornea (E, H, K) and the upper eyelid (L). Tbx3 was expressed only in the peripheral temporal retina (F). Tbx5 was expressed at high levels in a punctate pattern within the developing GCL throughout the dorsal hemisphere (G, J, M). (N–P) Adult eye sections of central retina close to the optic nerve. Tbx2 was expressed within the INL and the GCL (O). Tbx3 was expressed within the INL (P). Scale bar, 200 μm.

Table 1.

 

T-Box Gene Expression in the Developing Human and Mouse Retina

Human Gene	Weeks after Conception
		6	9	12	15
TBX2	D	inb	GCL, INL	GCL, INL
TBX3	D	D	D, INL	INL
TBX5	D	D	D, GCL	—

Mouse Gene	Embryonic Day
		10.5	12.5	14.5	Adult
Tbx2	D	D	D, INB	GCL, INL
Tbx3	D	D	D	INL
Tbx5	D	D	D, GCL	—

 D, expression within dorsal neural retina. Where expression is restricted to specific retinal layers, this is indicated as follows: INB, inner neuroblastic retina; GCL, ganglion cell layer; INL, inner nuclear layer.

The authors thank Gert Pflugfelder for the omb cDNA, Virginia Papaioannou for Tbx2, Tbx3, and Tbx5 cDNAs, Christine Campbell for TBX2 primers, and Angela Gouge for retinal cDNA; and Lesley Wong, Medical Research Council Tissue Bank, London, Philip Luthert, and Adam Rutherford for invaluable discussions and technical assistance.