|Home | About | Journals | Submit | Contact Us | Français|
Tbx2 is a T-box transcription factor gene that is dynamically expressed in the presumptive retina during optic vesicle invagination. Several findings implicate Tbx2 in cell cycle regulation, including its overexpression in tumours and regulation of proliferation during heart development. We investigated the role of Tbx2 in optic cup formation by analysing mice with a targeted homozygous mutation in Tbx2. Loss of Tbx2 caused a reduced presumptive retinal volume due to increased apoptosis, and a delay in ventral optic vesicle invagination leading to the formation of small and abnormally shaped optic cups. Tbx2 is essential for maintenance, but not induction of expression of the dorsal retinal determinant, Tbx5, and acts downstream of Bmp4, a dorsally expressed gene implicated in human microphthalmia. The small retina showed a hypocellular ventral region, loss of Fgf15, normally expressed in proliferating central retinal cells, and increased numbers of mitotic cells in the dorsal region, indicating that Tbx2 is required for normal growth and development across the D-V axis. Dorsal expression of potential regulators of retinal growth, Cyp1b1 and Cx43, and the topographic guidance molecule ephrinb2, was increased, and intra-retinal axons were disorganised resulting in a failure of optic nerve formation. Our data provide evidence that Tbx2 is required for proper optic cup formation and plays a critical early role in regulating regional retinal growth and the acquisition of shape during optic vesicle invagination.
The developing embryonic eye undergoes a complex process of morphogenesis and is patterned by polarised gene expression along its dorso-ventral, naso-temporal and proximo-distal axes. At 9.5 days post coitus (dpc) in the mouse, the optic vesicle forms as an extension of the forebrain neuroepithelium and invaginates upon contact with the surface ectoderm. Invagination initiates in the dorsal optic vesicle and gives rise to an optic cup, consisting of an expanded inner layer, the presumptive neural retina, and a thin outer layer, the presumptive retinal pigmented epithelium (Chow and Lang, 2001). Initially, the dorsal region of the forming optic cup is larger and developmentally more advanced, while formation of the ventral region of the optic cup lags in invagination, expansion, and neurogenesis (Coulombre et al., 1965; Romanoff, 1960). This results in an asymmetrically shaped optic cup by 11.5 dpc, that has an opening across the ventral surface, termed the optic fissure. The site at which future retinal ganglion cell (RGC) axons exit from the eye, termed the optic disc, is located ventrally at the proximal end of the optic fissure. By 13.5 dpc, upon closure of the optic fissure, the optic cup appears spherical with the optic disc positioned centrally. As a possible driving force for this shape alteration, the dorsal neural retina has been shown to possess a lower proliferative index compared to the ventral neural retina in the early optic cup, which later levels out (Behesti et al., 2006; Calvente et al., 1988; Morcillo et al., 2006). It is currently not known which genes regulate differential rates of cell division and contribute to the control of optic vesicle to optic cup morphogenesis.
Patterning of the D-V axis by asymmetrical gene expression is known to be critical for correct topographic mapping of retinal RGC axons from the neural retina to the brain (Koshiba-Takeuchi et al., 2000; Mui et al., 2002; Schulte et al., 1999). The idea that early D-V patterning may also be important for optic vesicle invagination and growth is supported by reports which show that mutations in the ventrally expressed homeobox genes, PAX2/Pax2, Vax2, and Vax1, cause coloboma, an eye malformation associated with blindness due to failure of optic fissure closure (Barbeiri et al., 2002; Bertuzzi et al., 1999; Eccles and Schimmenti, 1999; Mui et al., 2002; Torres et al., 1996). By contrast, little is known about the role of dorsally expressed genes. The earliest genes to be specifically expressed in the dorsal region of the developing eye include members of the evolutionarily conserved T-box transcription factor gene family (Bollag et al., 1994; Chapman et al., 1996; Sowden et al., 2001). Tbx2, Tbx3, and Tbx5, which share a high degree of amino acid sequence homology to the Drosophila optomoter blind (omb) gene, the ortholog of Tbx2 and Tbx3 (Papaioannou, 2001) are expressed in the distal layer of the optic vesicle and later in a nested fashion in the dorsal neural retina of the early optic cup (Behesti et al., 2006; Sowden et al., 2001). omb null mutants display abnormal optomotor behaviour due to the absence of lobula-plate giant nerve fibres, causing visual impairment (Pflugfelder et al., 1992). Misexpression of Tbx5 by electroporation in the developing chick eye leads to retinotectal misprojections, the expansion of the dorsally expressed ephrinB1 and ephrinB2 genes into the ventral neural retina, and the repression of the ventral genes, cVax and Pax2 (Koshiba-Takeuchi et al., 2000). Injection of synthetic Tbx2 or Tbx3 RNA into 2-cell stage Xenopus embryos has been reported to cause loss of ventral RPE pigmentation, medial displacement of the optic cup, and fused eyes in the most severe cases (Takabatake et al., 2002). In the zebrafish embryo, morpholino-mediated knock-down of Tbx2 resulted in a transient block of retinal cell types other than the RGCs, specifically in the dorsal neural retina (Gross and Dowling, 2005). However, the effect of loss of T-box gene function in mammalian eye development has not been investigated.
Current knowledge of Tbx2 points towards a role in cell cycle control and tissue growth. As well as being amplified in a subset of primary human breast tumours, in pancreatic tumour cell lines, and being overexpressed in melanoma cell lines (Bärlund et al., 2000; Jacobs et al., 2000; Mahlamaki et al., 2002; Sinclair et al., 2002; Vance et al., 2005), Tbx2 has been shown to function as an anti-senescence factor in vitro (Carlson et al., 2001; Jacobs et al., 2000; Vance et al., 2005). Furthermore, it has been shown to repress the p14ARF (Cdkn2a) promoter and those of other cyclin-dependant kinase inhibitors such as p21 in vitro (Lingbeek et al., 2002; Prince et al., 2004). These data, combined with the observation that Tbx2 expression is tightly regulated during the cell cycle have led to the proposal that in development Tbx2 may coordinate proliferation and cell cycle exit (Bilican and Goding, 2006). In vivo, during development, evidence for a role for Tbx2 in cell cycle control or tissue growth is currently sparse. However, it interacts with the closely related Tbx3 in maintenance and growth of the fetal mammary glands (Jerome-Majewska et al., 2005). In the developing heart, ectopic Tbx2 represses the expression of Nmyc in mice with a targeted deletion of Tbx20 (Cai et al., 2005).
We recently showed that Bmp4, a member of the TGFβ superfamily, regulates expression of Tbx2 in the early optic cup (Behesti et al., 2006), while mutation of the human BMP4 gene underlies anophthalmia, microphthalmia and brain anomalies in two families (Bakrania et al., 2008) and deletion of Bmp4 in mice disrupts lens and optic cup development (Furuta and Hogan, 1998). Although it has been reported that Tbx2 acts both as an activator and a repressor of gene expression depending on promoter context (Carreira et al., 1998; Habets et al., 2002; Paxton et al., 2002), the transcriptional activity and downstream targets have not been identified in eye development. Connexin 43 (Cx43), a gap-junction protein, is a candidate target of Tbx2, as it is co-expressed with Tbx2 in several developing tissues, including the chamber myocardium of the heart, the coronal sutures, and in the eye (Borke et al., 2003; Janssen-Bienhold et al., 1998; Yancey et al., 1992). The Cx43 promoter contains T-sites, the consensus T-box transcription factor binding sequence, and is repressed by Tbx2 in transfection assays (Borke et al., 2003; Chen et al., 2004).
To address the role of Tbx2 in mammalian optic cup formation, we analysed the phenotype of Tbx2tm1Pa/Tbx2tm1Pa(Tbx2-/-) homozygous mutant embryos. These mice carry a targeted deletion of part of the T-box binding domain. They die by 14.5 dpc presumably due to cardiac defects (Harrelson et al., 2004). We explored the hypotheses that (i) Tbx2 regulates optic vesicle invagination and regionalised cell proliferation during formation of the optic cup, and (ii) Tbx2, acting downstream of Bmp4, regulates the expression of other D-V patterning genes. We found that Tbx2-/- embryos show abnormal optic vesicle morphogenesis with a delay affecting the formation of the ventral region of the optic cup. The mutant optic cups display several abnormal features. The mutant presumptive neural retinas are smaller than in wild-type controls due to increased apoptosis, and show altered expression of proliferation markers phospho-histone H3 and Fgf15, indicating that Tbx2 is necessary for normal regional growth of the invaginating optic vesicle. Later they show disturbances in RGC axonal projections towards the optic disc. We found that Tbx2 is not required for induction, but is required for the maintenance of the expression domains of several D-V patterning genes that are critical for normal eye development.
Heterozygous timed matings were set up to generate wild-type and homozygous embryos for the Tbx2tm1Pa allele on a mixed (129/C57BL/ICR) genetic background. 47 pregnant females yielded a total of 363 embryos including 75 live embryos homozygous for the mutated Tbx2 allele. Dying or dead embryos, which displayed pericardial sac oedema and no heart beat or circulation at the time of dissection, were not included in the analysis. Embryos were genotyped from yolk sac DNA as previously described (Harrelson et al., 2004). Homozygous Chx10orJ/Chx10orJ embryos were obtained from timed matings between Chx10orJ/Chx10orJ homozygous mice on the 129/Sv background, while wild-type embryos were obtained from matings between 129/Sv mice. Embryos were dissected in phosphate buffered saline (PBS) pH 7.4 and fixed overnight in 4% w/v paraformaldehyde (Sigma, UK) in diethylpyrocarbonate (DEPC)-treated PBS at 4°C before processing for in situ hybridisation, immunohistochemistry and retinal volume analysis. For the histological analysis, embryos were fixed in Bouin's fixative and then embedded in paraffin. Transverse sections (8 μm) were prepared and stained with Haemotoxylin and Eosin (H&E).
To quantify the early optic cup size phenotype, retinal volume was estimated from measurements of retinal area on vibratome sections. Embryos were in embedded in a mix of 0.45% w/v gelatin, 28% w/v egg albumin, 18% w/v sucrose in PBS (all from Sigma, UK) and the block was hardened with 2.5% glutaraldehyde (Sigma, UK) at room temperature for one hour. Vibratome sections of 50 μm thickness were prepared using a Vibratome series 1000 sectioning system (Agar scientific Ltd, UK). Sections were mounted with 50% glycerol in PBS. Images were digitally captured on an Axiophot 2 microscope (Zeiss, Germany) fitted with Differential Interference Contrast (DIC) objectives and a Leica DC500 camera (Leica, Germany). Only specimens with a complete set of sections through the eye were included. Retinal area was measured on each section using OpenLab software and retinal volume obtained by multiplying retinal area by the thickness of the section and adding the values of sections per eye. Head length was measured using OpenLab software on images of whole mount embryos captured on a Leica MZ FLIII microscope (Leica, Germany) and a Leica DC500 camera (Leica, Germany). To assess generalised growth for each embryo, the length from the anterior groove of the forebrain to the posterior-most point of the hindbrain (roof of the fourth ventricle) was measured on the left and the right sides and the values averaged per embryo. Data were compared between mutant and wild-type embryos with the independent samples t-test.
Whole mount in situ hybridisation was performed on wild-type and mutant embryos stage matched by somite number using standard protocols and the following probes: Tbx2, Tbx3, Tbx5 (Behesti et al., 2006), Bmp4 (Furuta et al., 1997), Msx2 (1000 bp in pCR2.1, kindly provided by P. Sharpe, King's College, London), ephrinB2 (Efnb2; 3800 bp in pBSSK, kindly provided by D. Wilkinson, NIMR, London), Cyp1b1 (Stoilov et al., 2004), Vax2 (Schulte et al., 1999), Silver (Baxter and Pavan, 2003), Fgf15 (McWhirter et al., 1997), Jagged-1 (Mitsiadis et al., 1997).
Frozen sections (12 μm thick) were prepared from wild-type and mutant somite-matched embryos. Sections were blocked in 10% v/v heat-inactivated FCS (Sigma, UK), 1% BSA (Sigma, UK), 0.1% detergent. The primary antibodies used were as follows: phosphohistone H3 (pH3, 1:100, 0.1% Tween-20; Upstate, USA); Connexin43 (1:100, 0.1% Triton-X100, kindly provided by Dr D. Becker, UCL, (Wright et al., 2001); RMO270 (1:1500, 0.1% Triton-X100, Zymed, USA); VC1.1 (1:100, 0.1% Triton-X100, Sigma, UK). The secondary antibodies used were FITC-conjugated goat anti-mouse and FITC-conjugated goat anti-rabbit (1:100, Jackson Immunoresearch Labs, USA). The TUNEL assay (Roche, UK) was carried out according to manufacturer's description. Sections were counterstained with 1 μg/ml propidium iodide (Sigma, UK) and mounted with CITIFLUOR (Citifluor Ltd., UK).
Quantification of the relative levels of Connexin43 immunohistochemistry was performed using ImageJ (NIH) on blinded digital images of sections from age matched wild type (n=2) and mutant (n=3) embryos captured under identical exposure and contrast settings using a fluorescent Axiophot 2 microscope (Zeiss, Germany) and OpenLab software.
Sections were divided into a dorsal and a ventral region as demarcated in Fig. 3E for 10.5 dpc retina. The presumptive neural retinal region lying in close proximity to the surface ectoderm was analysed for 9.5 dpc sections. Mitotic indices and total cell number were calculated per section and the data were checked for normal distribution. The number of pH3-positive mitotic cells per section (14 sections from 6 mutant eyes and 8 sections from 4 wild-type eyes at 10.5 dpc, and 25 sections from 6 mutant eyes and 32 sections from 8 wild-type eyes at 9.5 dpc) in each region were analysed with a one-way ANOVA, while the mitotic indices were analysed with an ANCOVA with the total number of cells per region or per section included as a covariate. The number of apoptotic nuclei were counted on 25 sections from 8 wild-type eyes and 24 sections from 6 mutant eyes at 9.5 dpc, and 16 sections from 4 wild-type eyes and 24 sections from 6 mutant eyes at 10.5 dpc and the TUNEL index was compared with the Mann-Whitney test for nonparametric data.
In wild type embryos Tbx2 expression is first detected at 9.5 dpc in the distal neuroepithelium of the optic vesicle, in close proximity to the overlying surface ectoderm, extending along the length of the dorso-ventral (D-V) axis (Fig. 1A, B). Tbx2 is then restricted to the dorsal region of the presumptive neural retina during invagination of the optic vesicle (Behesti et al., 2006). By 11.5 dpc, Tbx2 expression is confined to the dorsal retina in the developing optic cup (Fig. 1C, D). To assess the requirement for Tbx2 in optic vesicle invagination and growth, Tbx2-/- embryos were analysed between 9.5-12.5 dpc. Amongst the live embryos obtained from heterozygous matings, 20% (75/363) were homozygous for the mutated Tbx2 allele, which is consistent with the previously described embryonic survival rate of homozygous embryos (Harrelson et al., 2004).
Tbx2-/- embryos displayed normal optic vesicle formation (data not shown). However by 10.5 and 11.5 dpc, the invaginating optic vesicle and the optic cup were smaller in mutants than in wild-type littermates (Fig. 1G-J). Lens vesicle invagination initiated normally but a marked reduction in the size of the optic cup and lens vesicle was apparent by 11.5 dpc (Fig. 1E-H, K, L). The subretinal space in the naso-ventral optic cup often appeared abnormally large in mutants, and the RPE pigmentation was delayed compared to the wild type (Fig. 1G, H). Although optic vesicle invagination was underway at 10.5 dpc in the dorsal region (Fig. 1I, J), it was delayed in the ventral region in mutants compared to somite-matched wild-type littermates (Figs. 3A-D, dashed lines and 4E, F).
By 11.5 and 12.5 dpc, the optic cup was markedly shorter along the proximo-distal (P-Di) axis (Fig. 1K-N), indicating delayed morphogenesis and growth of the neural retina, RPE and lens. Moreover at 12.5 dpc, the eye lid grooves, which can normally be distinguished at this stage in wild-type embryos (Fig. 1O, arrows) had not formed in mutants (Fig. 1P). The mutants also displayed an unusually dense hyaloid vasculature, a small lens vesicle (Fig. 1K, L), and primitive primary lens fibre elongation compared with advanced elongation in stage-matched wild-type embryos (Fig. 1M, N). Finally, the nerve fibre layer and optic nerve were not distinguishable in mutants on H&E stained sections, while fibres were present and exited the optic cup in wild-type littermates at 12.5 dpc (Fig. 1M, N). The severity of these phenotypes was variable in mutant embryos. At 12.5 dpc, the full phenotype was observed in 8/12 eyes, whereas 4/12 eyes manifested a milder phenotype showing only reduced eye size and a slight decrease in the P-Di axial length (data not shown). Mutant eyes with the severe phenotype protruded much less than wild-type counterparts (Fig. 1O, P, compare solid black lines). These data indicate that targeted disruption of Tbx2 results in delayed growth and invagination of the presumptive neural retina of the optic vesicle leading to abnormal development of the optic cup. The abnormalities observed by E12.5 affect several tissues external to the site of Tbx2 expression and are likely a downstream consequence of the loss of Tbx2 during optic vesicle invagination.
To quantify the early effect of loss of Tbx2, the presumptive retinal volume was compared between mutant and wild-type embryos at somite stages (ss) 33-35 (10.5 dpc). The optic vesicle and the lens vesicle invaginate between ss 26 and 33 (Behesti et al., 2006). Measurements of head length, used as a measure of growth during early embryogenesis (Brown et al., 1990), served to test for a generalised growth delay in mutants. Mutant presumptive retinae were significantly smaller in volume than wild-type retinae (p= 0.007; Fig. 2A). By contrast, head length measurements did not differ significantly between mutant and wild-type embryos (p= 0.352, Fig. 2B). These data indicate that the lack of Tbx2 causes an early small retina phenotype that does not correlate with a general developmental delay. Reduced retinal volume, largely due to failure of development of the ventral retina was apparent as early as E10.5, prior to closure of the lens vesicle (Fig. 1I, J; 4E, F).
A comparative analysis of cell number in the dorsal and ventral retina (as demarcated in Fig. 3E) was performed on sections of mutant and wild-type embryos. This confirmed that the ventral retina is indeed hypocellular by 10.5 dpc whereas the dorsal retina was unaffected (Fig. 2C). These data show that Tbx2 expression is required for normal growth across the D-V axis during optic vesicle invagination.
To examine a role for apoptosis in causing the small retina phenotype in Tbx2-/- embryos the TUNEL assay, which detects apoptosis, was performed on somite-matched embryos at ss 30-31 (10.5 dpc) during optic vesicle invagination, when the ventral retina first showed signs of underdevelopment. TUNEL labelling revealed a normal pattern of cell death in mutants that resembled previously published apoptosis data in the 10.5 dpc mouse retina (Trousse et al., 2001) (Fig. 3A, B; Table 1, P= 0.358).
We next examined cell division in Tbx2-/- mutants by pH3 immuno-labelling on somite-matched embryos. Labelled retinal progenitors in M-phase were correctly located at the ventricular surface of the neural retina (Sidman and Smelser, 1961; Dhomen et al., 2006) in both mutant and wild-type embryos (Fig. 3C, D). Comparisons of pH3 cell counts and mitotic indices (pH3 cells/total cells) throughout the neural retina (global) showed a trend of an increase in the numbers of mitotic cells in mutants compared to wild types at ss 30-31 (Table 1; Fig. 3C, D; P= 0.10). Cell division in the dorsal retina only was also compared. Significantly more cells were in M-phase in mutants compared to wild-type embryos in the dorsal neural retina (Fig. 3E; P= 0.04). However, there was no difference in the mitotic index of the ventral retina (data not shown).
As the increased level of dorsal mitotic cells does not readily provide an explanation for the delayed optic vesicle invagination and hypocellular ventral retina (Fig. 2C), we also performed TUNEL and pH3 analysis at an earlier stage of development at ss 20-24 (9.5 dpc) prior to optic vesicle invagination. At this stage, Tbx2 expression is not yet restricted to the dorsal retina, and is instead expressed in both the dorsal and ventral regions of the presumptive retina of the optic vesicle (Fig. 1B). A significantly increased level of TUNEL labelling index was detected in the mutant retina compared with somite-matched wild types at 9.5 dpc (Fig. 4A, B, C).
In contrast to the reduced retinal cell number observed one day later, at this earlier stage, an increased overall cell number was observed, (Fig. 4D). There were also significantly more mitotic cells present in the mutant presumptive retina compared to wild types at 9.5 dpc (Fig. 4E). Thus, in the absence of Tbx2, an increased mitotic index was observed at sites normally expressing Tbx2, namely the distal optic vesicle at 9.5 dpc and the dorsal neural retina at 10.5 dpc. An increased number of pH3 positive cells were also detected in the ventral mesenchyme between the hypoplastic ventral neural retina and the surface ectoderm, where Tbx2 expression is normally observed (Fig. 1D, arrowhead and Fig. 3C, D, arrowheads).
Together these data show that loss of Tbx2 at optic vesicle stage, leads to increased cell death and dysregulation of cell proliferation. These changes occur prior to a visible morphological abnormality in the mutants and coincide with the earliest phase of Tbx2 expression in wild types. Hence, cell death at 9.5 dpc causes the small retinal volume observed by 10.5dpc, and altered regional cell proliferation and cell death cause the abnormal optic vesicle morphogenesis observed by E10.5. The hypocellularity of the ventral retina therefore results from the failure of normal growth and invagination across the D/V axis. As reduced lens vesicle size (Figs. 1L and 3B, D) was also observed, non cell autonomous signalling dependent on Tbx2 expression must be required for growth of the lens tissue.
Fgf15 has been implicated in dorsal midbrain growth (Ishibashi and McMahon, 2002; Saitsu et al., 2005). It is first expressed, like Tbx2, in the distal optic vesicle at 9.5 dpc (Kurose et al., 2004; McWhirter et al., 1997) and then as Tbx2 expression is restricted to the dorsal neural retina, Fgf15 continues to be expressed in dividing retinal progenitor cells in the optic cup during retinal histogenesis (Blackshaw et al., 2004; Rowan et al., 2004). Moreover, its reduced expression in two mouse models, the Chx10orJ/orJ ocular retardation mutant (Rowan et al., 2004) and Bmpr1a-/-/Bmpr1b-/- mutants (Murali et al., 2005) correlates with a failure of normal optic cup growth. Notably, Bmp4-/- optic vesicles also showed loss of Fgf15 expression, that could be restored by application of exogenous BMP4 (Murali et al., 2005). At 10.5 dpc in wild types, Fgf15 and Tbx2 showed complementary expression patterns in central and dorsal presumptive retina respectively (Fig. 5A, C). In the Tbx2 mutants we found that at early optic vesicle stage, Fgf15 was correctly expressed (data not shown), whereas after invagination at ss 35, Fgf15 expression was absent in 4/6 mutant retina and weakly expressed in the other two (n= 4/6 eyes; Fig. 5A, B and data not shown). The loss of Fgf15 expression supports the conclusion that Tbx2 is required for normal regional proliferation across the D-V axis.
Analysis of Tbx2 expression in Chx10orJ/orJ mutants, revealed that loss of Chx10 did not affect Tbx2 expression (n= 6/6 eyes; Fig. 5C, D) suggesting that Tbx2 is not regulated by Chx10. In Chx10orJ/orJ mutants that lack Chx10 function, pigment genes normally expressed only in the RPE of the optic cup are expanded into the hypoplastic neural retina (Burmeister et al., 1996; Horsford et al., 2004; Rowan et al., 2004). To investigate whether a similar loss of retinal specification linked to reduced proliferation, was occurring in Tbx2-/- embryos, we examined the expression of the pigment gene, Silver (Baxter and Pavan, 2003). In contrast to the Chx10orJ/orJ mutants, in the Tbx2-/- optic cups, Silver expression was similar to that of wild-type embryos and was restricted to the RPE (Fig. 5 E, F), suggesting that the neural retina/RPE boundary is not compromised in Tbx2-/- embryos. Together these data indicate that while retinal specification occurs without Tbx2, it is required for regulation of proliferating progenitors during formation of the optic cup.
Cx43 is a putative downstream target of Tbx2 in other tissues and has also been implicated in retinal growth (Becker and Mobbs, 1999; Kjaer et al., 2004; Paznekas et al., 2003; Vitiello et al., 2005). CX43 mutation in humans results in oculo-dento-digital dysplasia, which is characterised by microphthalmia, cataract, and abnormalities of the iris, together with syndactyly and craniofacial skeletal defects, (Kjaer et al., 2004; Paznekas et al., 2003; Vitiello et al., 2005). Furthermore, it has been shown that the RPE, a site of Cx43 expression, regulates proliferation in the underlying neural retina by the release of ATP through gap junction Cx43 hemichannels (Pearson et al., 2005a; Pearson et al., 2005b). Immunohistochemistry in somite-matched Tbx2-/- mutant and wild-type retinae at 10.5 dpc during optic vesicle invagination showed Cx43 protein was distributed in the dorsal RPE, but not in the neural retina. However, in mutants the Cx43 expression domain appeared increased in the dorsal, nasal, and temporal RPE, compared to weak or barely detectable labelling in equivalent regions in wild-type eyes (n= 6/6 eyes; wild type = 74.3 +/- 28.7; mutant =112 +/- 24.7 fluorescent intensity units; Fig. 5 G-J). Therefore, while the appropriate expression of Silver indicates that RPE specification is not affected by the loss of Tbx2, the upregulated Cx43 expression in the RPE of mutant eyes shows that Tbx2 plays a role in mediating interactions between the neural retina and RPE during optic vesicle morphogenesis that are known to be important for the regulation of proliferation. Altered Cx43 expression across the D-V axis provides a possible mechanism to regulate regional proliferation by modulating ATP release.
Previously we identified Tbx2 as a downstream target of Bmp4 in the developing optic cup (Behesti et al., 2006). Bmp4 is thought to be a primary mediator of dorsal specific gene expression (Koshiba-Takeuchi et al., 2000) and its mutation in humans was recently shown to cause a spectrum of eye abnormalities including microphthalmia (Bakrania et al., 2008). Here we investigate whether Tbx2, acting downstream of Bmp4, is responsible for the regulation of other D-V patterning genes. Whole mount in situ hybridisation was carried out at 10.5-11.5 dpc of the dorsal markers Tbx3, Tbx5, Msx2, ephrinB2 and Cyp1b1, and the ventral marker Vax2.
We first investigated whether the population of cells in which Tbx2 would normally be expressed are present in their normal location in the mutants. Labelling of cells using a full-length Tbx2 probe, which recognizes the mutant RNA, confirmed that lack of Tbx2 does not result in death nor in migration of these cells from the dorsal optic cup (n= 4/4 eyes, Fig. 6 A, B). The expression of Bmp4 in the dorsal neural retina was also normal at both 10.5 and 11.5 dpc (Fig. 6C-F), indicating that Tbx2 does not feed back into the regulation of Bmp4 expression in the optic cup.
The Bmp-responsive genes Msx2, Tbx3, and Tbx5 were all expressed dorsally in the developing eye of Tbx2-/- embryos, however, although Msx2 and Tbx3 expression was normal in mutants (n= 4/4 eyes per probe; Fig. 6G-J), the expression domain of Tbx5 was markedly reduced in size in comparison to somite-matched wild-type littermates at 10.5 and 11.5 dpc (n= 6/6 eyes; Fig. 6K-N). In contrast to the reduction in Tbx5 expression dorsally, ephrinB2 appeared expanded to cover a wider region of the dorsal optic cup (n= 5/6 eyes; Fig. 6O, P). Furthermore, the expression of Cyp1b1 in the dorsal-most region of the neural retina appeared expanded as compared with stage-matched wild-type littermates (n= 5/6 eyes; Fig. 6Q, R). We also observed that while Vax2 expression was induced normally in the ventral retina of the Tbx2-/- embryos, by 10.5 dpc the Vax2 expression domain in the ventral optic cup was reduced (n= 6/6 eyes; Fig. 6S, T), correlating with the reduced size of the ventral neural retina (Figs. 3A-D, dashed lines and 5E, F and and2C).2C). Together, these analyses show that Tbx2 is not required for the onset of expression of these D-V patterning genes and hence D-V polarisation of the optic vesicle. Rather, Tbx2 is required to maintain correct spatial expression of the key dorsal retinal determinant, Tbx5, during optic cup formation. In addition, Tbx2 is needed to delimit the normal dorsal expression domain of the topographic guidance molecule ephrinb2 and the retinoic acid (RA) generating enzyme, Cyp1b1 (Chambers et al., 2007). The fact that these dorsally expressed genes are affected differently by loss of Tbx2 (Tbx5 reduced, ephrinB2 and Cyp1b1 increased, Msx2, Bmp4 and Tbx3 unchanged) argues against the observed changes simply reflecting the altered patterns of growth occurring without Tbx2. All of these downstream genes are known to be critically important for eye development. Mutations in the human CYP1B1 gene cause primary congenital glaucoma (Stoilov et al., 1997); MIM601771) whereas the Tbx5 and ephrinB2 genes are implicated in D-V patterning, axon guidance and topographic mapping of RGCs from the retina to the brain.
Finally we examined whether loss of Tbx2 and the delayed ventral morphogenesis and altered gene expression patterns observed by 10.5dpc in the Tbx2 mutants affected the early stages of retinal differentiation in the optic cup. Fibroblast growth factor signalling in the central neural retina is known to trigger retinal cell differentiation in chick and zebrafish embryos (Martinez-Morales et al., 2005) and we observed loss of Fgf15 expression in the Tbx2-/- optic cups. Moreover, the altered expression of Tbx5 and ephrinB2 in the mutant eyes may affect RGC axonal projections. Immunohistochemistry was carried out on sections of 12.5 dpc mutant and wild-type littermate eyes with the RMO270 and the VC1.1 antibodies. RMO270 labels RGCs from the point of becoming post-mitotic at the ventricular surface of the neural retina, allowing detection of newly born cells migrating to the vitreal surface and their axonal projections (McCabe et al., 1999). VC1.1 detects amacrine and horizontal cells, and at 12.5 dpc labels precursors destined to differentiate into amacrine and horizontal cells (Alexiades and Cepko, 1997).
RMO270 labeling revealed that RGC differentiation in the centro-dorsal retina was underway in both mutant and wild types. Normally the central retina differentiates first and the peripheral region last (Dräger, 1985), and this pattern was maintained in the mutants (Fig. 7A, B), therefore excluding premature differentiation as a factor contributing to the small optic cup phenotype. In both genotypes, RGCs migrated to the correct location adjacent to the vitreous. By 12.5 dpc in wild-type eyes an organised nerve fibre layer was present in the central neural retina, dorsal to the optic disc (Fig. 7A). A similar spatial pattern of labelling was detected in mutant eyes (n= 4 eyes; Fig. 7B), though in contrast to wild-type eyes, no axons projected through the poorly defined optic disc (Fig. 7E, F), consistent with observations on H&E sections (Fig. 1M, N). In comparison to the neat layer of RGC axons in wild-type retinae, axons were extremely disorganised in mutant retinae, where the majority appeared to project into the vitreous chamber rather than follow the vitreal surface towards the optic nerve head (Fig. 7C-F).
VC1.1 immunohistochemistry at 12.5 dpc labelled cell bodies and their processes in the central neural retina in a similar pattern in both mutant and wild-type eyes (n= 4; Fig. 7G-L), suggesting that commitment to amacrine and horizontal cell fates also occurs normally in Tbx2-/- mutants. Together these data suggest that the onset of retinal histogenesis and the migration of cells to adopt a correct position across the laminar axis of the developing retina are not affected by the loss of Tbx2. However, intraretinal RGC axon guidance and optic nerve development appear impaired.
We show that without Tbx2, presumptive neural retinal volume is reduced and invagination of the ventral optic vesicle is delayed resulting in the formation of small and abnormally shaped optic cups in Tbx2-/- mutants. This is the first report to demonstrate that Tbx2 is required for proper optic cup formation. Investigation of premature cell differentiation, and expansion of the RPE at the expense of the neural retina, excluded these mechanisms as the cause of the small optic cup phenotype. Instead we observed increased cell death, and an abnormal increase in the number of mitotic cells, in the mutant retina, coincident with the dynamic change in Tbx2 expression. We also found abnormal expression of Cx43, a known regulator of retinal proliferation, as well as a loss of expression of the retinal progenitor cell marker Fgf15 in the central retina. These findings support the conclusion that Tbx2 acts to modulate apoptosis and regional levels of proliferation across the presumptive neural retina as its expression is restricted dorsally during optic vesicle to optic cup transition.
Tbx2 is initially expressed across the D-V axis in the distal optic vesicle. Our data indicate that this early phase of Tbx2 expression followed by development of the high dorsal retinal expression of Tbx2 is needed for coordination of growth and morphogenesis across the D-V axis (Fig. 8). Without Tbx2 expression in the distal optic vesicle increased cell death occurs prior to initiation of invagination causing a hypocellular ventral retina apparent one day later. Restriction of Tbx2 to the developmentally-advanced dorsal retinal region during invagination is essential to suppress dorsal growth and this action appears to be required for the acquisition of the globe shape of the optic cup. Our data therefore indicates that Tbx2 regulates growth and morphogenesis by modulating levels of apoptosis and facilitating regional differences in levels of proliferation across the presumptive neural retina.
Several in vitro studies have shown that Tbx2 acts to promote cell cycle progression/cell division, although clear evidence for a similar role during development has not been found. Instead, analysis of the developing heart (Cai et al., 2005; Ribeiro et al., 2007) and eye, point to an opposite role, where Tbx2 slows down regional cell proliferation. In zebrafish, the pattern of proliferation along the looping heart tube is dynamic and changes from a pattern of homogenous proliferation, to a heterogeneous one in which dividing cells are more concentrated in the future chambers. This process appears to be regulated by Tbx3 and Tbx2, as overexpression caused a decrease in the number of dividing cells, whereas fish embryos injected with morpholino against tbx3b and/or tbx2a showed an increased number of dividing cells, and in both cases regionalisation of proliferating cells was lost (Ribeiro et al., 2007). Recently, the Tbx2 ortholog omb was proposed to act as a repressor of retinal growth acting to delimit the extent of the eye field in Drosophila. omb mutants display an increase in ommatidial number, while ectopic over expression of omb caused reduced eye size, or even the complete loss of retinal tissue (Porsch et al., 2005). Our data are in agreement with this emerging role of Tbx2 as a regulator of regional growth during development and these findings highlight the importance of controlling cell proliferation as a driving force of morphogenesis.
Several of the phenotypes observed in Tbx2-/- embryos are similar to those observed in retinoic acid (RA) deficient embryos. Certainly aspects of the Tbx2-/- phenotype are consistent with Tbx2 regulation of growth factor signalling that influences surrounding tissue non cell autonomously. RA deficiency in zebrafish embryos results in lack of ventral neural retina formation (Marsh-Armstrong et al., 1994) and in mammals causes microphthalmia and anophthalmia (Kalter and Warkany, 1959), while retroviral expression of a dominant negative form of the RA receptor α (RARα) in the chick results in a small eye phenotype (Sen et al., 2005). Other remarkably similar abnormalities were recently detected in Aldh1a1; Aldh1a3 compound null mutants (which lack retinaldehyde dehydogenase activity), including shortening of the ventral neural retina, expansion of the ventral extra-ocular mesenchyme and failure of eye lid groove formation (Molotkov et al., 2006). This latter study concluded that RA sources external to the optic cup, guide eye morphogenesis via paracrine signalling but are unnecessary for retinal dorsoventral patterning as both Tbx5 expression and Vax2 expression were unaffected (Molotkov et al., 2006). Notably, we found that loss of Tbx2 caused an increase in dorsal Cyp1b1 expression. CYP1B1 synthesises both all-trans-retinal and all-trans-RA independently of the major RA synthesising enzymes (RALDH1-3), and contributes to D-V patterning in the chick neural tube (Chambers et al., 2007). Whether increased CYP1B1 activity affects retinal growth is not known. However, the similarities in these phenotypes suggest that investigation of the interplay between Tbx2 and the network of enzymes and receptors that modulate RA signalling, will offer further insight into the relationships between these pathways in optic cup formation.
Bmp4, expressed in the dorsal optic cup, is essential for the expression of several other dorsally expressed genes including Tbx2 (Behesti et al., 2006) and the repression of the ventrally expressed gene Vax2 (Koshiba-Takeuchi et al., 2000; Murali et al., 2005). Here we have shown that Tbx2 is not required for Bmp4 expression, nor for the onset of expression of several other D-V patterning genes. Instead, it is required for the maintenance of Tbx5, and the restriction of ephrinB2 and Cyp1b1 expression in the dorsal optic cup. Whereas chick Tbx5 misexpression ectopically induced ephrinB2 expression (Koshiba-Takeuchi et al., 2000), our genetic analysis showed that changes in ephrinB2 expression can occur independently of Tbx5 in the mouse. In fact ephrinB2 expanded ventrally in Tbx2-/- mutants while Tbx5 was reduced. Furthermore, although Bmp4 is essential for the induction of Tbx5 expression (Murali et al., 2005) and can induce ectopic Tbx5 expression (Behesti et al., 2006; Koshiba-Takeuchi et al., 2000) here we show reduced Tbx5 expression while Bmp4 expression is unaffected. Together, these data indicate that Tbx2 is required to maintain Tbx5 expression in the dorsal retina, whereas the other Bmp4-responsive genes, Msx2 and Tbx3 are unaffected by loss of Tbx2. Our study clearly positions Tbx2 as a downstream effector of Bmp4 signalling; loss of Tbx5 and Fgf15 expression is seen in both Bmp4-/- (Murali et al., 2005) and Tbx2-/- optic cups, whereas Bmp4 expression is unchanged in the Tbx2 mutants. However, the fact that Vax2 expression expands into the dorsal retina in the Bmp4-/- optic cup (Murali et al., 2005), whereas the ventral Vax2 expression domain is reduced in the Tbx2 mutant, shows that Tbx2 functions not to induce, or to repress Vax2, rather it is needed for generation of the ventral tissue that expresses Vax2.
Functional studies have previously shown that Tbx5 and ephrinB2 expression domains are required for D-V patterning and topographic mapping of RGC axons from the dorsal neural retina to the brain (Koshiba-Takeuchi et al., 2000; Mann et al., 2002) and here we show that disruption of these gene expression domains is associated with intraretinal RGC axon guidance problems in Tbx2 mutant optic cups. In our study the simplest explanation for the RGC guidance problems is that they are secondary to the patterning and morphological defects in the optic cup, although a direct requirement for Tbx2 in RGCs has not been excluded. Significantly, in the chick following Vax2 (cVax) misexpression, dorsal and ventral RGC axons exhibited defects in intraretinal pathfinding near the optic disc (Muhleisen et al., 2006). A number of other different sets of positive and negative factors cooperate in the process of intraretinal guidance (Bao, 2008). These include BMP7, possibly secreted by the periocular mesenchyme, which is needed for the generation of the optic disc and normal RGC routing (Morcillo et al., 2006). However, loss of Bmp7 was associated with reduced apoptosis and normal Tbx5 expression, both not seen in the Tbx2 mutants, suggesting distinct mechanisms.
Morpholino-mediated knock-down of Tbx2b in the zebrafish embryo resulted in a temporary block to differentiation only in the dorsal retina (Gross and Dowling, 2005). Markers of amacrine, and rod and cone photoreceptors cells were downregulated in the zebrafish dorsal retina, whereas RGCs were less affected and axons projected to the optic disc to form the optic nerve (Gross and Dowling, 2005). Here, our findings on retinal differentiation in the Tbx2-/- mouse are contrary to the zebrafish knockdown, although this may be due to the later stages of retinal development examined in the zebrafish. We provide evidence for the existence of a dorsal differentiating cell population in the Tbx2-/- mouse retina, by immunolabelling with VC1.1 which labels precursors of amacrine and horizontal cells (Alexiades and Cepko, 1997) and by labelling for RGCs which show a normal spatial onset of differentiation. Therefore in the mouse, dorsal retinal progenitors are, at least initially, fated towards distinct cell types.
We propose a model for the function of Tbx2 during optic cup formation (Fig. 8) in which Tbx2 acts downstream of the dorsal determinant Bmp4 and is required for modulating retinal growth during optic vesicle invagination. Tbx2 controls both cell death and regionalised cell proliferation, and is essential to maintain expression of Tbx5 and Fgf15 across the D-V axis, and to restrict dorsal expression of Cx43, Cyp1b1 and ephrinb2. Tbx2 is initially expressed in the entire distal optic vesicle, where it represses proliferation. As Tbx2 expression becomes restricted to the dorsal presumptive neural retina, it slows growth regionally, allowing the ventral retinal region to proliferate more and “catch-up” in terms of growth with the dorsal neural retina. This coordination of growth across the D-V axis is required for normal morphogenesis of the optic vesicle to form the optic cup.
The study was supported by grants from the Child Health Research Appeal Trust, a Bogue Research Fellowship (H. B.), the Ulverscroft Foundation, and NIH grant HD033082 (V. E. P). Research at the Institute of Child Health and Great Ormond Street Hospital for Children NHS Trust benefits from R&D funding received from the NHS Executive. We are very grateful to B. Hogan, P. Sharpe, C. Cepko, I. Stoilov, D. Wilkinson, T. Wright, and U. Lendahl for providing cDNAs used to prepare in situ hybridisation probes, D. Becker for providing Cx43 antibody and E. Ernstoff and C. A. Schall for assistance in collecting embryos.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.