Ординатура / Офтальмология / Английские материалы / Eye, Retina, and Visual System of the Mouse_Chalupa, Williams_2008

Figure 25.4 Ectopic horizontal cell–rod synapses in the absence of Rb. A, Horizontal cells are among the first cell types produced in the developing retina. However, two of their synaptic partners, rods and bipolar cells, are among the last cells produced during development. B, In the absence of Rb in the Chx10-Cre;RbLox/Lox retinas, there are mosaic patches in the ONL lacking rod photoreceptors. This is a unique phenotype because these cells do not adopt another cell fate and remain relatively quiescent. This is an

disruption of developing horizontal processes in the OPL of Rb-deficient retinas was evident at P7, when photoreceptor terminals are just beginning to appear. Apically projecting horizontal cell neurites persisted in Chx10-Cre;RbLox/− retinas, but in control littermates, horizontal processes reorganized into their lateral orientation. Although these studies are useful for identifying changes in synaptogenesis during retinal development, they provide only snapshots of individual stages. Ideally, one should follow maturation and synapse formation of individual horizontal cells throughout retinogenesis to determine whether the apical horizontal cell processes reflect a defect in the normal developmental process or active reorganization of normal lateral horizontal cell process in response to the failure of rods to form in the

Chx10-Cre;RbLox/Lox retinas.

Another important consideration is the role of Rb in horizontal cell development. It is possible that Rb is required for proper horizontal cell differentiation and that the defect in horizontal cell synaptogenesis is due to the absence of Rb in horizontal cells rather than the absence of their rod photoreceptor synaptic partner. As discussed in this chapter, elucidating the cell-autonomous and non-cell-autonomous roles of Rb in retinal development can best be studied using genetic mosaic analysis. The Chx10-Cre;RbLox/Lox;Z/AP mouse is ideally suited for this type of analysis. Alkaline phosphatase (AP) is expressed in every cell lacking Rb. By comparing the contribution of AP+ and AP− horizontal cell processes to ectopic synapses using TEM of lead citrate–stained sections of Chx10-Cre;RbLox/Lox;Z/AP retinas, we can deter-

ideal model in which to explore the process of synaptogenesis between rods, bipolar cells, and horizontal cells in the absence of direct synaptic contacts apical to the developing INL cells. C, In the absence of rod inputs, horizontal cells extend their processes deep into the ONL and form ectopic synapses with normal rods in neighboring mosaic patches of retina. There are no bipolar dendrites in these ectopic synapses.

mine whether Rb is required cell autonomously in developing horizontal cells for proper differentiation and synaptogenesis.

Summary

Until relatively recently, it was assumed that cell cycle genes regulated proliferation during development and that the multiple family members for each gene simply reflected functional redundancy for this essential regulatory network. However, the studies discussed in this chapter emphasize that cell cycle proteins have roles that extend far beyond that of proliferation control, and that individual family members can have both overlapping and unique functions. The Rb family is an excellent example of these principles. Both Rb and p107 regulate RPC proliferation, and they appear to be somewhat interchangeable in this process. Beyond cell cycle regulation, Rb also plays a role in regulating rod photoreceptor development, and p107 cannot substitute for this unique role of Rb in rod photoreceptors. In the absence of Rb, rods fail to form and remain as immature cells in the ONL. The mosaic nature of Rb inactivation using Chx10Cre;RbLox/Lox mice led to important advances in our understanding of synaptogenesis between rods, horizontal cells, and bipolar cells. Moreover, genetic studies of single and compound knockout mice for the Rb family shed light on the complex redundant and compensatory mechanisms that prevent deregulated proliferation of RPCs during development. These data were used to generate the first knockout

mouse models of retinoblastoma and to test new treatments in preclinical models, and they have had direct effects on clinical trials for this debilitating childhood cancer. Both developmental biology and cancer biology benefit when these two traditionally separate fields are brought together in the study of retinal development and of diseases of the retina.

REFERENCES

Akey, D. T., Zhu, X., Dyer, M., Li, A., Sorensen, A., Blackshaw, Fukuda-Kamitani, T., Daiger, S. P., Craft, C. M., et al. (2002). The inherited blindness associated protein AIPL1 interacts with the cell cycle regulator protein NUB1. Hum. Mol. Genet. 11(22):2723–2733.

Aslanian, A., Iaquinta, P. J., Verona, R., and Lees, J. A. (2004). Repression of the Arf tumor suppressor by E2F3 is required for normal cell cycle kinetics. Genes Dev. 18(12):1413–1422.

Burmeister, M., Novak, J., Liang, M. Y., Basu, S., Ploder, L., Hawes, N. L., Vidgen, D., Hoover, F., Goldman, D., et al. (1996). Ocular retardation mouse caused by Chx10 homeobox null allele: Impaired retinal progenitor proliferation and bipolar cell differentiation. Nat. Genet. 112(4):376–384.

Cepko, C. L., Austin, C. P., Yang, X., Alexiades, M., and Ezzeddine, D. (1996). Cell fate determination in the vertebrate retina.

Proc. Natl. Acad. Sci. U.S.A. 93(2):589–595.

Chen, D., Livne-Bar, I., Vanderluit, J. L., Slack, R. S., Agochiya, M., and Bremner, R. (2004). Cell-specific effects of RB or RB/p107 loss on retinal development implicate an intrinsically death-resistant cell-of-origin in retinoblastoma. Cancer Cell 5(6):539–551.

Chen, S., and Dyer, M. (2008). [Identification of Rb target genes in the retina.] Unpublished results.

Clarke, A. R., Maandag, E. R., van Roon, M., van der Lugt, N. M., van der Valk, M., Hooper, M. L., Berns, A., and te Riele, H. (1992). Requirement for a functional Rb-1 gene in murine development. Nature 359(6393):328–330.

Coles, B. L., Angenieux, B., Inoue, T., Del Rio-Tsonis,K., Spence, J. R., McInnes, R. R., Arsenijevic, Y., and van der Kooy, D. (2004). Facile isolation and the characterization of human retinal stem cells. Proc. Natl. Acad. Sci. U.S.A. 101(44):15772–15777.

Dakubo, G. D., Wang, Y. P., Mazerolle, C., Campsall, K., McMahon, A. P., and Wallace, V. A. (2003). Retinal ganglion cell-derived sonic hedgehog signaling is required for optic disc and stalk neuroepithelial cell development. Development 130(13):2967–2980.

Donovan, S. L., and Dyer, M. A. (2004). Developmental defects in Rb-deficient retinae. Vision Res. 44(28):3323–3333.

Donovan, S. L., and Dyer, M. A. (2005). Regulation of proliferation during central nervous system development. Semin. Cell Dev. Biol. 16(3):407–421.

Donovan, S. L., Schweers, B., Martins, R., Johnson, D., and Dyer, M. A. (2006). Compensation by tumor suppressor genes during retinal development in mice and humans. BMC Biol. 4:14.

Dyer, M. A. (2003). Regulation of proliferation, cell fate specification and differentiation by the homeodomain proteins Prox1, Six3, and Chx10 in the developing retina. Cell Cycle 2(4): 350–357.

Dyer, M. A. (2004). Mouse models of childhood cancer of the nervous system. J. Clin. Pathol. 57(6):561–576.

Dyer, M. A., and Bremner, R. (2005). The search for the retinoblastoma cell of origin. Nat. Rev. Cancer 5(2):91–101.

Dyer, M. A., and Cepko, C. L. (2000a). Control of Müller glial cell proliferation and activation following retinal injury. Nat. Neurosci. 3(9):873–880.

Dyer, M. A., and Cepko, C. L. (2000b). p57(Kip2) regulates progenitor cell proliferation and amacrine interneuron development in the mouse retina. Development 127(16):3593–3605.

Dyer, M. A., and Cepko, C. L. (2001a). p27Kip1 and p57Kip2 regulate proliferation in distinct retinal progenitor cell populations. J. Neurosci. 21:4259–4271.

Dyer, M. A., and Cepko, C. L. (2001b). The p57(Kip2) cyclin kinase inhibitor is expressed by a restricted set of amacrine cells in the rodent retina. J. Comp. Neurol. 429(4):601–614.

Dyer, M. A., and Cepko, C. L. (2001c). Regulating proliferation during retinal development. Nat. Rev. Neurosci. 2(5):333– 342.

Dyer, M. A., and Harbour, J. W. (2006). Cellular events in tumorigenesis. In A. D. Singh et al. (Eds.), Clinical ocular oncology. London: Elsevier.

Dyer, M. A., Livesey, F. J., Cepko, C. L., and Oliver, G. (2003). Prox1 function controls progenitor cell proliferation and horizontal cell genesis in the mammalian retina. Nat. Genet. 34(1):53–58.

Dyer, M. A., Rodriguez-Galindo, C., and Wilson, M. W. (2005). Use of preclinical models to improve treatment of retinoblastoma. PLoS Med. 2(10):e332.

Friend, S. H., Bernards, R., Rogelj, S., Weinberg, R. A., Rapaport, J. M., Albert, D. M., and Dryja, T. P. (1986). A human DNA segment with properties of the gene that predisposes to retinoblastoma and osteosarcoma. Nature 323(6089):643– 646.

Furukawa, T., Morrow, E. M., and Cepko, C. L. (1997). Crx, a novel otx-like homeobox gene, shows photoreceptor-specific expression and regulates photoreceptor differentiation. Cell 91(4):531–541.

Gustincich, S., Feigenspan, A., Wu, D. K., Koopman, K. L., and Raviola, E. (1997). Control of dopamine release in the retina: A transgenic approach to neural networks. Neuron 18(5):723– 736.

Haider, N. B., Naggert, J. K., and Nishina, P. M. (2001). Excess cone cell proliferation due to lack of a functional NR2E3 causes retinal dysplasia and degeneration in rd7/rd7 mice. Hum. Mol. Genet. 10(16):1619–1626.

Hinds, J. W., and Hinds, P. L. (1979). Differentiation of photoreceptors and horizontal cells in the embryonic mouse retina: An electron microscopic, serial section analysis. J. Comp. Neurol. 187(3):495–511.

Howes, K. A., Ransom, N., Papermaster, D. S., Lasudry, J. G., Albert, D. M., and Windle, J. J. (1994). Apoptosis or retinoblastoma: Alternative fates of photoreceptors expressing the HPV-16 E7 gene in the presence or absence of p53. Genes Dev. 8(11):1300–1310.

Jacks, T., Fazeli, A., Schmitt, E. M., Bronson, R. T., Goodell, M. A., and Weinberg, R. A. (1992). Effects of an Rb mutation in the mouse. Nature 359(6393):295–300.

Jeon, C. J., Strettoi, E., and Masland, R. H. (1998). The major cell populations of the mouse retina. J. Neurosci. 18(21): 8936–8946.

Johnson, D. A., Donovan, S. L., and Dyer, M. A. (2006). Mosaic deletion of Rb arrests rod differentiation and stimulates ectopic synaptogenesis in the mouse retina. J. Comp. Neurol. 498(1): 112–128.

Jones, B. W., Watt, C. B., Frederick, J. M., Baehr, W., Chen, C. K., Levine, E. M., Milam, A. H., Lavail, M. M., and Marc, R. E. (2003). Retinal remodeling triggered by photoreceptor degenerations. J. Comp. Neurol. 464(1):1–16.

Knudson, A. (1971). Mutation and cancer: Statistical study of retinoblastoma. Proc. Natl. Acad. Sci. 68:820–823.

Laurie, N. A., Donovan, S. L., Shih, C. S., Zhang, J., Mills, N., Fuller, C., Teunisse, A., Lam, S., Ramos, Y., et al. (2006). Inactivation of the p53 pathway in retinoblastoma. Nature 444 (7115):61–66.

Laurie, N. A., Gray, J. K., Zhang, J., Leggas, M., Relling, M., Egorin, M., Stewart, C., and Dyer, M. A. (2005). Topotecan combination chemotherapy in two new rodent models of retinoblastoma. Clin. Cancer Res. 11(20):7569–7578.

Lees, E., Faha, B., Dulic, V., Reed, S. I., and Harlow, E. (1992). Cyclin E/cdk2 and cyclin A/cdk2 kinases associate with p107 and E2F in a temporally distinct manner. Genes Dev. 6(10): 1874–1885.

Lobe, C. G., Koop, K. E., Kreppner, W., Lomeli, H., Gertsenstein, M., and Nagy, A. (1999). Z/AP, a double reporter for Cre-mediated recombination. Dev. Biol. 208(2):281–292.

Ma, C., Papermaster, D., and Cepko, C. L. (1998). A unique pattern of photoreceptor degeneration in cyclin D1 mutant mice.

Proc. Natl. Acad. Sci. U.S.A. 95(17):9938–9943.

MacLaren, R. E., Pearson, R. A., MacNeil, A., Douglas, R. H., Salt, T. E., Akimoto, M., Swaroop, A., Sowden, J. C., and Ali, R. R. (2006). Retinal repair by transplantation of photoreceptor precursors. Nature 444(7116):203–207.

MacPherson, D., and Dyer, M. A. (2007). Retinoblastoma: From the two-hit hypothesis to targeted chemotherapy. Cancer Res. 67(16):7547–7550.

MacPherson, D., Sage, J., Kim, T., Ho, D., McLaughlin, M. E., and Jacks, T. (2004). Cell type-specific effects of Rb deletion in the murine retina. Genes Dev. 18(14):1681–1694.

Marc, R. E., and Jones, B. W. (2003). Retinal remodeling in inherited photoreceptor degenerations. Mol. Neurobiol. 28(2): 139–147.

Marc, R. E., Jones, B. W., Watt, C. B., and Strettoi, E. (2003). Neural remodeling in retinal degeneration. Prog. Retin. Eye Res. 22(5):607–655.

Marino, S., Vooijs, M., van Der Gulden, H., Jonkers, J., and Berns, A. (2000). Induction of medulloblastomas in p53-null mutant mice by somatic inactivation of Rb in the external granular layer cells of the cerebellum. Genes Dev. 14(8):994–1004.

Martins, R. A., Linden, R., and Dyer, M. A. (2006). Glutamate regulates retinal progenitors cells proliferation during development. Eur. J. Neurosci. 24(4):969–980.

McArdle, C. B., Dowling, J. E., and Masland, R. H. (1977). Development of outer segments and synapses in the rabbit retina. J. Comp. Neurol. 175(3):253–274.

Mears, A. J., Kondo, M., Swain, P. K., Takada, Y., Bush, R. A., Saunders, T. L., Sieving, P. A., and Swaroop, A. (2001). Nrl is required for rod photoreceptor development. Nat. Genet. 29(4): 447–452.

Redburn, D. A., and Madtes, P. (1986). Postnatal development of 3H-GABA-accumulating cells in rabbit retina. J. Comp. Neurol. 243(1):41–57.

Robanus-Maandag, E., Dekker, M.,van der Valk, M., Carrozza, M. L., Jeanny, J. C., Dannenberg, J. H., Berns, A., and te Riele, H. (1998). p107 is a suppressor of retinoblastoma development in pRb-deficient mice. Genes Dev. 12(11):1599– 1609.

Rowan, S., and Cepko, C. L. (2004). Genetic analysis of the homeodomain transcription factor Chx10 in the retina using a novel multifunctional BAC transgenic mouse reporter. Dev Biol. 271(2):388–402.

Schweers, B. A., and Dyer, M. A. (2005). Perspective: New genetic tools for studying retinal development and disease. Vis. Neurosci. 22(5):553–560.

Sharma, R. K., O’Leary, T. E., Fields, C. M., and Johnson, D. A. (2003). Development of the outer retina in the mouse. Brain Res. Dev. Brain Res. 145(1):93–105.

Skapek, S. X., Lin, S. C., Jablonski, M. M., McKeller, R. N., Tan, M., Hu, N., and Lee, E. Y. (2001). Persistent expression of cyclin D1 disrupts normal photoreceptor differentiation and retina development. Oncogene 20(46):6742–6751.

Sun, H., Chang, Y., Schweers, B., Dyer, M. A., Zhang, X., Hayward, S. W., and Goodrich, D. W. (2006). An E2F bindingdeficient Rb1 protein partially rescues developmental defects associated with Rb1 nullizygosity. Mol. Cell Biol. 26(4):1527– 1537.

Tropepe, V., Coles, B. L., Chiasson, B. J., Horsford, D. J., Elia, A. J., McInnes, R. R., and van der Kooy, D. (2000). Retinal stem cells in the adult mammalian eye. Science 287(5460): 2032–2036.

Vooijs, M., te Riele, H., van der Valk, M., and Berns, A. (2002). Tumor formation in mice with somatic inactivation of the retinoblastoma gene in interphotoreceptor retinol binding proteinexpressing cells. Oncogene 21(30):4635–4645.

Zhang, H. S., Gavin, M., Dahiya, A., Postigo, A. A., Ma, D., Luo, R. X., Harbour, J. W., and Dean, D. C. (2000). Exit from G1 and S phase of the cell cycle is regulated by repressor complexes containing HDAC-Rb-hSWI/SNF and Rb-hSWI/SNF. Cell 101(1):79–89.

Zhang, J., Gray, J., Wu, L., Leone, G., Rowan, S., Cepko, C. L., Zhu, X., Craft, C. M., and Dyer, M. A. (2004a). Rb regulates proliferation and rod photoreceptor development in the mouse retina. Nat. Genet. 36(4):351–360.

Zhang, J., Schweers, B., and Dyer, M. A. (2004b). The first knockout mouse model of retinoblastoma. Cell Cycle 3(7):952– 959.

Zhang, H., Xiong, Y., and Beach, D. (1993). Proliferating cell nuclear antigen and p21 are components of multiple cell cycle kinase complexes. Mol. Biol. Cell 4(9):897–906.

Zhu, C. C., et al. (2002). Six-3 mediated auto repression and eye development requires interaction with members of the Grouch-related family of co-repressors. Development 129(12): 2835–2849.

The mammalian retina is an exquisitely patterned neural tissue that arises from an amorphous sheet of dividing neuroblasts during fetal and neonatal life (figure 26.1A). By determining the mechanisms that underlie the formation of the retina, we can improve our understanding of CNS development as well as establish more robust methods for the repair and regeneration of damaged retinas (for a recent example, see MacLaren et al., 2006). Impressive advances have been made over the past decade that elucidate many of the crucial features controlling retinal development.

In the broadest sense, two developmental processes work in parallel to construct the retina, cellular lamination and cell type differentiation (figures 26.1B and C). Contrary to expectation, these processes may be genetically separable (Fu et al., 2006). In this chapter we concentrate primarily on the mechanisms that control the differentiation of retinal cell types, with particular emphasis on retinal ganglion cells (RGCs). In all vertebrates, differentiation of the six retinal neurons (rod, cone, horizontal, bipolar, amacrine, and ganglion) and one glial cell type (Müller) proceeds in a sequential manner, with RGCs the first cells to differentiate and Müller glia the last (for a recent review, see Cayouette et al., 2006). The retinal cell types form as subpopulations of retinal progenitor cells (RPCs) progress through a classical developmental sequence of cell competence, specification/commitment, and differentiation.

The differentiation of each retinal cell type is a highly complex process requiring both extrinsic and intrinsic factors. Recent investigations have emphasized the importance of distinct subpopulations of RPCs, each programmed to generate a limited number of cell types (Mu et al., 2005b; Yan et al., 2005; Cayouette et al., 2006). These subpopulations can easily be observed in the differential expression of a host of transcription factors, which act in combination to control specific retinal cell fates (Furukawa et al., 1997; Brown et al., 1998; Mears et al., 2001; Hatakeyama and Kageyama, 2004; Mu and Klein, 2004; Ohtoshi et al., 2004; Li et al., 2004; Mu et al., 2005a; Feng et al., 2006; Fujitani et al., 2006). Our intention here is to review recent informa-

tion on the intrinsic genetic program that regulates RGC formation. Studies on RGC development in the mouse and other vertebrates are beginning to reveal the detailed behavior of RPCs during the early stages of retinogenesis. Consequently, we can now follow RGC progression from naive, uncommitted neuroblasts to their ultimate fate as fully differentiated neurons capable of receiving, processing, and transmitting the electrical signals required in the brain for visual perception.

A gene regulatory network for retinal ganglion cell development

To better conceptualize the regulatory events we wish to describe, we rely on a gene regulatory network model for RGC development that we have previously proposed (Mu et al., 2005a) and continue to develop (figure 26.2). It is becoming increasingly apparent that studying the functions of groups of interacting molecules offers novel, more global ways of understanding developmental processes than does studying linear pathways or single elements (see Nature 435:1, 2005). Establishing comprehensive networks of interacting molecules can explain emerging behaviors in ways that cannot be done by studies of linear pathways or single elements within a network. Our model is based on analogous gene regulatory networks constructed by Davidson and co-workers for sea urchin development (Davidson et al., 2002, 2003; Amore et al., 2003). In all these models, cis-regulatory elements associated with each gene in the network are identified and connected to their corresponding transcription factors, which in turn serve to regulate gene expression through cis-regulatory elements on genes encoding other transcription factors. In our RGC gene regulatory network, most cis-regulatory elements are predicted to exist based largely on circumstantial evidence, and their existence still requires direct confirmation. Nonetheless, as can be seen in figure 26.2, the RGC gene regulatory network features a downward cascade of transcription factors occupying at least four different tiers. This

Figure 26.1 Developing and mature mouse retinas. A, An embryonic retina at E12.5. B, A postnatal retina at P16. Nuclei are labeled in red with propidium iodide and represent the positions of the cell soma. Neurite processes are labeled in green and blue with antineurofilament antibody and anticholine acetyltransferase antibody, respectively, and represent axons and dendrites. C, Schematic representation of the mature retina. RGCs and displaced

transcription factor cascade leads ultimately to the expression of downstream genes whose products are required to form a mature, functional RGC. The network is constructed on the basis of results obtained from gene expression profiling, spatiotemporal expression patterns, chromatin immunoprecipitation analyses, and retinal phenotypes associated with various mouse knockout lines (Mu et al., 2005a, 2008).

Early regulatory events in retinogenesis

The retina originates in embryogenesis from the optic vesicle, which forms at the anteriormost portion of the neural tube. For normal eye development to occur, extensive interactions are required between the overlying surface ectoderm (the presumptive lens) and the evaginating optic vesicle. Following the allocation of the optic vesicle into the future retinal pigmented epithelium and the prospective neural retina, a number of important transcription factors are expressed in the neural retina that are required for establishing a field of actively dividing, multipotent RPCs. These so-called panneural retinal determination factors include Rx/Rax, which is essential for the formation of RPCs (Zhang et al., 2000; Bailey et al., 2004), Prox1, Six3, and Chx10, which have roles in regulating RPC proliferation and cell fate specifica-

amacrine cells are found within the ganglion cell layer (GCL). Intermediate neurons (horizontal, bipolar, and amacrine) are found within the inner nuclear layer (INL), and rod and cone photoreceptor cells are found within the outer nuclear layer (ONL). IPL and OPL are inner plexiform and outer plexiform layers where axons and dendrites synapse. RGC, retinal ganglion cell. See color plate 14.

tion (Burmeister et al., 1996; Dyer, 2003), and Sox2, a dose-dependent regulator of RPC competence required for Notch1 signaling (Taranova et al., 2006). Retinas from mice harboring either germ line or retina-specific null mutations in the genes encoding any of the previously mentioned transcription factors have severe defects that affect the entire RPC population. This indicates that these factors are positioned upstream of the RGC gene regulatory network and represent genes at the top of the regulatory hierarchy for retinogenesis (see figure 26.2). Despite the strict requirement for these and several other transcription factors in controlling various aspects of RPC proliferation and competence, their precise functions and relationships to each other remain unclear. Nevertheless, it is becoming apparent that a complex retinal determination gene network involving extensive feedback and autoregulatory circuitry underlies the expression of these regulatory genes in the early developing retina and sits above the RGC gene regulatory network (Marquardt, 2003; Silver and Rebay, 2005).

A key retinal determination factor expressed in the early developing retina is Pax6, a pair-rule homeobox-containing transcription factor that is essential for the formation of all retinal cell types with the exception of amacrine cells (Marquardt et al., 2001; Marquardt, 2003). RPCs expressing

Figure 26.2 A gene regulatory network model for RGC development. Genes are depicted such that the right side of the bent arrow for each gene indicates the gene product and the left side indicates the transcriptional control region. Solid lines connecting genes indicate established upstream-downstream relationships, and dashed lines suggest inferred relationships. Downward-pointing arrows indicate gene activation, and downward perpendicular lines indicate gene repression. In most cases the connections have not been shown to be direct. Genes boxed in yellow represent regulators of general retinal competence and RPC proliferation; genes boxed in orange are proneural bHLH genes associated with establishing competence in

Pax6 represent the first step in the progression toward RGC commitment, although it is clear that Pax6 has broader roles in retinogenesis (Marquardt and Gruss, 2002). Evidence gathered from several vertebrate species indicates that the Notch-Delta signaling pathway, mediated by the basic helix- loop-helix (bHLH) transcription factors Hes1 and Hes5, is essential for maintaining Pax6-expressing cells in a proliferative state and inhibiting the differentiation of retinal cell types (Perron and Harris, 2000; Jadhav et al., 2006a). Recent studies suggest that Notch1 not only maintains the progenitor cell state by lateral inhibition, as was previously thought, but is also required to inhibit photoreceptor cell fate; in

RPCs for specific retinal cell fates. The light brown-purple box represents the general repression of RGC genes hypothesized to be mediated by the neural differentiation transcriptional repressor NSRF/REST (Mu et al., 2005b); the blue and purple-blue boxes represent genes encoding RGC-specific upstream (blue) and downstream (purple-blue) transcription factors. Genes boxed in green are those that encode proteins associated with RGC maturation and function. Genes boxed in purple encode secreted signaling molecules. In some cases only representative examples are shown for each box. A more detailed description can be found in Mu et al., 2005b. See color plate 15.

retinas where Notch1 is removed early in retinogenesis, cone photoreceptor cells differentiate at the expense of other retinal cell types ( Jadhav et al., 2006b; Yaron et al., 2006). This provocative result suggests that suppression of cone photoreceptor cell fate by Notch1 allows for the specification of the other retinal cell types (Yaron et al., 2006). Whatever the precise mechanism of Notch1 signaling, it seems clear that an early step in retinogenesis is the loss of Notch1 activity in a subpopulation of RPCs destined to become RGCs (Nelson et al., 2006). The basis on which this subpopulation of RPCs is selectively released from the Notch signal is unclear, but in the absence of Notch1 or Hes1, the intrinsic

genetic program required for RGC development is initiated ( Jadhav et al., 2006a; Nelson et al., 2006). RPCs in this subpopulation thus acquire the competence for committing to an RGC fate and are the first RPCs in the retina to undergo this transition. As retinogenesis proceeds, Notch1 activity is overridden in other RPC subpopulations, thereby leading to the differentiation of the later retinal cell types (Perron and Harris, 2000; Jadhav et al., 2006a).

Math5 and retinal ganglion cell formation

Math5 and Retinal Ganglion Cell Competence RPCs become competent to form RGCs by activating the expression of the gene encoding the proneural bHLH transcription factor Math5, a mouse orthologue of Drosophila Atonal (Brown et al., 1998; Vetter and Brown, 2001; Le et al., 2006). Math5 expression is first detected at embryonic day 11 (E11) in RPCs distributed throughout the neuroblast layer. Gene knockouts have demonstrated that in the absence of Math5, RGCs do not develop (Brown et al., 2001; Kay et al., 2001; Wang et al., 2001). Pax6-Math5-expressing RPCs are believed to define a competence field that permits the subsequent steps of RGC formation to proceed (Wang et al., 2001; Yang et al., 2003). Although expression of Math5 is necessary, it is not sufficient for RGCs to form. Math5 is expressed in a larger number of RPCs than would be expected if its sole function were dedicated to RGC development (Wang et al., 2001). Moreover, lineage-tracing experiments indicate that Math5- expressing cells can give rise to other retinal cell types besides RGCs (Yang et al., 2003).

Results obtained from mice, Xenopus, chicks, and zebrafish have demonstrated the importance of Math5 (Ath5 in Xenopus, chick, and zebrafish) in vertebrate retinogenesis (reviewed in Vetter and Brown, 2001). Recent reports have identified a number of downstream genes whose expression is dependent on the presence of Math5 or Ath5 orthologues, including some genes that may be direct targets (Yang et al., 2003; Logan et al., 2005; Mu et al., 2005a). Identifying genes regulated by Math5 has provided valuable information for constructing the RGC gene regulatory network with Math5 positioned as the central node (see figure 26.2). Although understanding the role of Math5 in the developing retina clearly requires further study, a picture is now emerging on how this proneural bHLH transcription factor acts to establish neuronal cell competence.

Math5 and Cell Cycle Progression Math5 has at least three critical functions in RPCs of the developing retina. It promotes cell cycle progression, suppresses non-RGC fates, and activates RGC fate. Math5 and Ath5 orthologues are thought to promote cell cycle progression by causing RPCs to exit the cell cycle and advance to G0 (Kay et al., 2001;

Yang et al., 2003; Matter-Sadzinski et al., 2005; Le et al., 2006). Time-lapse analysis in zebrafish retinas demonstrates a striking correlation in the expression of zebrafish Ath5, cell cycle exit, and RGC differentiation (Poggi et al., 2005). In this study, RPCs expressing green fluorescent protein (GFP) under the control of the Ath5 gene promoter were monitored by three-dimensional time-lapse analysis. Ath5-expressing cells divide just once along the circumferential axis to produce two postmitotic daughters, one of which becomes an RGC. However, in Ath5 mutant retinas, which lack RGCs, Ath5-expressing RPCs divide along the centralperipheral axis and produce two RGCs. These results suggest that extrinsic signals act on Ath5-expressing RPCs to influence the orientation of cell division and alter normal cell fate (Poggi et al., 2005). Brown and co-workers have suggested a possible mechanism for cell cycle exit (Le et al., 2006). These investigators have shown that Math5 is required for the expression of the cell cycle inhibitor p27/Kip1; RPCs lacking Math5 have diminished and shifted levels of p27/Kip1 and are unable to appropriately exit the cell cycle (Le et al., 2006). Currently, it is not known whether Math5 regulates p27/Kip1 directly or indirectly. It is also not clear whether Math5 is solely responsible for the cell cycle exit of Math5expressing cells.

Math5 and Suppression of Non–Retinal Ganglion Cell

Fates Math5 also plays a critical role in suppressing the development of other retinal cell types by inhibiting the expression of the genes encoding NeuroD1, Math3, Neurogenin2 (Ngn2), and Bhlhb5, all of which are bHLH transcription factors that play important roles in the formation of rod and cone photoreceptor cells, amacrine cells, and bipolar cells (Hatakeyama and Kageyama, 2004; Mu et al., 2005b; Feng et al., 2006; Le et al., 2006). NeuroD1 and Ngn2 are expressed in mitotically active RPCs, and Ngn2 expression may initially overlap with that of Math5. Notably, in chick retinal development, Ngn2 and Ath5 compete to regulate the Ath5 promoter (Matter-Sadzinski et al., 2001, 2005). In chicks, Ngn2 is thought to be the initial activator of Ath5 expression following the downregulation of Hes1 (Matter-Sadzinski et al., 2005). Chick Ath5 autoregulates its own expression after initial activation by Ngn2 (MatterSadzinski et al., 2005), but this autoregulatory mechanism does not appear to operate for Math5 in the mouse retina (Hutcheson et al., 2005). The chick results suggest that once Ath5 is expressed, it represses Ngn2 expression. Because Ngn2 has the ability to activate Ath5, additional control mechanisms must be in place to ensure Ngn2 does not activate Ath5 inappropriately. It is apparent that during retinogenesis, RPC subpopulations distributed throughout the neuroblast layer are delicately balanced in ways that can be profoundly affected by altering the levels and combinations of bHLH and other transcription factors (Hatakeyama and

Kageyama, 2004; Mu et al., 2005a, 2005b). Thus, Hes1- expressing cells remain as multipotent, actively dividing RPCs, Math5-expressing cells exit the cell cycle and advance to an RGC competent state, and Neurod-, Math3-, Ngn2-, and Bhlhb5-expressing cells assume competence for other retinal cell fates.

Math5 and the Activation of Retinal Ganglion Cell

Fate Based on the retinal phenotypes of Math5 knockout mice, the most conspicuous role for Math5 is the activation of the genetic program responsible for RGC differentiation. The earliest manifestation of Math5’s action is the activation of four genes, each of which encodes a transcription factor that is first expressed as RGCs begin to differentiate. The four factors are POU domain factor Pou4f2 (also called Brn3b), LIM homeodomain factor Isl1, transcriptional repressor Groucho family member Tle1, and zinc finger protein Myt1 (see figure 26.2; Mu et al., 2005a). The expression of these factors is dependent on the presence of Math5, as revealed by gene expression profiling experiments that compared Math5 mutant and wild-type retinas (Mu et al., 2005a). Pou4f2 is the only transcription factor of the four thus far shown to be essential for RGC development. Retinas from Pou4f2 knockout mice have defects in RGC differentiation but not in the initial specification of RGCs (Gan et al., 1996, 1999; Erkman et al., 2000; Wang et al., 2000). This phenotype is consistent with the placement of Pou4f2 genetically downstream of Math5 (see figure 26.2).

Pou4f 2, Isl1, Tle1, and Myt1, and possibly other earlyexpressed transcription factors (Yang et al., 2003; Logan et al., 2005) control the expression of downstream genes that encode additional transcription factors expressed at later times in RGC development (see figure 26.2). Transcription factors whose expression depends on Pou4f2 have been identified by gene expression profiling that compared wildtype retinas with Pou4f 2 mutant retinas (Mu et al., 2004). These Pou4f2-dependent transcription factors and the additional transcription factors whose expression is independent of Pou4f2 represent the next tier in the RGC gene regulatory network. Math5 therefore exerts its function in RGC development largely by initiating a transcription factor cascade, mediated in part by Pou4f2, which ultimately leads to the activation of genes required for the formation of the mature RGC. Because not all Math5-expressing cells become committed to an RGC fate, additional factors must participate so that the RGC transcriptional cascade is activated in only a subset of Math5-expressing cells.

Regulation of Math5 expression

Activation of MATH5 in Retinal Progenitor Cells A critical but unresolved question in retinal development is

how different RPC subpopulations come to express distinct combinations of transcription factors. This complex process is made even more complicated by the fact that different RPC subpopulations intermingle and are distributed relatively uniformly throughout the neuroblast layer, appearing and disappearing as retinogenesis proceeds. Extrinsic factors are likely to have roles in establishing the developmental potential of RPCs, but only a few of these factors have been characterized (e.g., Harpavat and Cepko, 2003). It now seems likely that the extrinsic environment may not be the major driver for specifying RPC fates and that RPCs may be developmentally constrained by intrinsic programs set up at very early times in retinogenesis (Livesey and Cepko, 2001; Mu et al., 2005b; Cayouette et al., 2006). Because Math5 is essential for RGC formation and RGCs are the first retinal cell type to differentiate, understanding the intrinsic mechanisms that control the expression of Math5 is particularly important for developing a meaningful model of retinogenesis.

Math5 is activated in a subpopulation of RPCs as Hes1 and Hes5 expression is downregulated. Retinal-specific expression of reporter transgenes using 5′-regulatory DNA from Math5 or from Ath5 orthologues from mouse, Xenopus, zebrafish, and chicken has been reported (Matter-Sadzinski et al., 2001, 2005; Hutcheson et al., 2005; Poggi et al., 2005). Some of these reports implicate bHLH factor-binding sites (E-boxes) as cis-regulatory elements involved in controlling Ath5 expression. Matter-Sadzinski et al. (2001, 2005) have suggested that Ngn2 binds to an E-box within the chick Ath5 promoter to activate Ath5 expression. However, in the mouse, Ngn2’s involvement in Math5 expression has not been demonstrated. In addition, while Ath5 may autoregulate its expression in chicks (Matter-Sadzinski et al., 2001, 2005), Math5 expression does not appear to be autoregulatory in the mouse retina (Hutcheson et al., 2005). These results suggest that there are significant differences in the mechanisms that control Math5 and Ath5 expression among vertebrate species.

Besides E-boxes, additional cis-regulatory elements are likely to be required to activate Math5/Ath5 expression, but at present, these elements remain poorly defined. Pax6 must play some role in regulating Math5 expression in the retina because in the absence of Pax6, Math5 is not expressed (Marquardt et al., 2001). It is not clear, however, whether Pax6 is a direct activator of Math5. Conserved Pax6-binding sites have been found in the 5′-regulatory DNA of Math5 but it is not clear whether these sites are functional Pax6 regulatory elements (Nadean Brown, pers. comm.).

Regulating the Duration of MATH5 Expression In addition to the mechanisms that control the activation of Math5 expression, additional mechanisms must operate

Figure 26.3 Segregated expression of Math5 and Pou4f2 in retinogenesis. The image represents an E14.5 retina from a knock-in mouse in which a gene encoding an HA-epitope-tagged Math5 replaces the endogenous Math5 allele. Math5-HA expression is detected by immunostaining with an anti-HA antibody (red). Pou4f2 expression is detected by immunostaining with an antiPou4f2 antibody (green). Yellow staining shows the overlap in expression of Math5 and Pou4f2. See color plate 16.

to maintain Math5’s expression over a relatively narrow temporal and spatial window. This is because Math5 is transiently expressed during retinal development, with a peak between E12.5 and E15.5 (Brown et al., 1998; Kim et al., 2005; Poggi et al., 2005). A subset of Math5-expressing RPCs commit to an RGC fate by activating Pou4f2 and other early expressing transcription factor genes (Mu et al., 2005a). This commitment step correlates with a substantial downregulation in Math5 expression spatially (Brown et al., 1998; Poggi et al., 2005). Direct visualization of Pou4f2 activation and Math5 downregulation is depicted in figure 26.3, which shows an E14.5 retina doubly immunostained for Pou4f2 and HA-tagged Math5. In the neuroblast layer, many Math5-expressing cells are detected (figure 26.3). Math5-Pou4f2-expressing cells are also seen in the neuroblast layer, although the double-labeled cells tend to be concentrated to a zone more basally located than cells expressing just Math5 (see figure 26.3). In the emerging ganglion cell layer, Pou4f2-expressing cells are readily seen, but in this region, very few Math5-expressing cells or Math5-Pou4f2- expressing cells are detected. This result supports a model in which Math5 activates the expression of Pou4f2 in a subset of RPCs in the neuroblast layer. Shortly after Pou4f2 is activated, Math5 expression is extinguished as newly differentiating RGCs begin their migration to the ganglion cell layer.

The tightly regulated expression of Math5 may serve as an important mechanism to coordinate RGC development

with other retinal cell types. Recent results suggest that regulating the duration of Math5 expression is crucial in determining whether an RPC will commit to an RGC fate or some other fate (Kim et al., 2005), and this seems to be mediated by the TGF-β superfamily member GDF11. GDF11 controls the number of RGCs, as well as amacrine and photoreceptor cells that form during retinal development (Kim et al., 2005). GDF11 does not affect the proliferation of RPCs, as it does in other neuronal progenitor cells, but instead it controls the duration of Math5 expression; retinas lacking GDF11 have prolonged Math5 expression and produce an excess of RGCs at the expense of other retinal cell types (Kim et al., 2005). GDF11 is expressed in both RPCs and RGCs and appears to work by a feedback mechanism in which the fates of the multipotent RPCs expressing Math5 can be altered by modulating the duration of Math5 expression (Kim et al., 2005).

Repressing MATH5 Expression in Other Retinal Progenitor Cells Subpopulations of actively dividing, multipotent neuroblasts arise soon after the neural retina first emerges as a distinct neuroepithelial sheet. As discussed earlier, one of these subpopulations contains multipotent RPCs that activate the expression of Math5 and form a field of RPCs competent to become RGCs. However, other RPC subpopulations programmed to specify other retinal cell fates also arise at this time. These other subpopulations will never express Math5. A significant issue that still must be addressed centers on the mechanisms by which individual intrinsic programs are set up in distinct RPC subpopulations.

Recent progress has been made in addressing this issue by the identification of two transcription factors, Foxn4, a winged helix forkhead factor, and Ptf1a, a bHLH factor, which are both required to determine horizontal and amacrine cell fates during mouse retinogenesis (Li et al., 2004; Fujitani et al., 2006). Foxn4 knockout mice have revealed that Foxn4 controls the formation of amacrine and horizontal cells by activating the expression of Math3, NeuroD1, and Prox1, and that Foxn4 cooperates with Pax6 to perform its functions (Li et al., 2004). Loss of Foxn4 appears to result in an overproduction of rod photoreceptor cells (Li et al., 2004). Ptf1a knockout mice have revealed that Ptf1a marks postmitotic RPCs with competence to produce horizontal and amacrine cells exclusively (Fujitani et al., 2006). In this study, Ptf1a was identified as a primary downstream target for Foxn4. Notably, the loss of Ptf1a results in an upregulation of Pou4f2 and an increase in the number of RGCs (Fujitani et al., 2006). Together, the results of Li et al. (2004) and Fujitani et al. (2006) suggest a model for horizontal and amacrine cell formation in which RPCs expressing Pax6 and Foxn4 activate the expression of Ptf1a. RPCs in this subpopulation can take on one of two pathways; in one,

Prox1 is activated downstream of Ptf1a and horizontal cells are specified, and in the other, Math3 and NeuroD1 are activated in conjunction with Ptf1a, and amacrine cells are specified (Li et al., 2004; Fujitani et al., 2006).

Math5 is not expressed in the Foxn4-Ptf1a RPC lineage, and Foxn4 and Ptf1a are not expressed in Math5-expressing RPCs. This suggests that negative cross-regulation among these transcription factors is operating to keep Math5 repressed in Foxn4-Ptf1a-expressing RPCs and Foxn4 and Ptf1a repressed in Math5-expressing RPCs. Intriguingly, loss of Ptf1a results in upregulation of Pou4f2 (Fujitani et al., 2006), implying that Math5 is also upregulated in Ptf1a mutant retinas. Indeed, Ptf1a may directly repress Math5 expression or regulate other transcriptional activity that leads to Math5 repression. However, the fact that Foxn4-null retinas do not upregulate Pou4f2 is perplexing, since Foxn4 activates the expression of Ptf1a (Li et al., 2004; Fujitani et al., 2006). Ptf1a null retinas, however, continue to express Foxn4, which may indicate that Ptf1a null RPCs reflect an altered competence state that differs from that of Foxn4 null RPCs.

The identification of distinct RPC subpopulations for RGC and horizontal and amacrine cell competence highlights the importance of intrinsic programs that control the fates of RPCs early in retinogenesis. It is remarkable that in the developing retina, these distinct subpopulations are spatially interspersed and temporally overlapped. What remains a mystery is the mechanism that allows one RPC to assume an RGC-competent state while a nearby RPC takes on a horizontal or amacrine cell–competent state. It may be that the intrinsic programs for these RPC subpopulations are set up much earlier than is currently thought, perhaps even before the onset of retinogenesis.

Regulation of retinal ganglion cell differentiation downstream of Math5

Transition from Retinal Ganglion Cell Competence to Retinal Ganglion Cell Commitment The biological significance of Math5 downregulation in newly differentiating RGCs is not known. However, this downregulation is concurrent with the loss of multipotency and irreversible commitment to RGC differentiation. The regulatory mechanisms underlying RGC commitment are likely to be associated with what has been termed a lockdown state by Davidson and co-workers for irreversible differentiation in embryo development (Davidson et al., 2002, 2003; Levine and Davidson, 2005). Three steps are required for achieving a lockdown state in RGCs. First, Math5 must activate the genes encoding the early-expressing transcription factors Pou4f2, Isl1, Tle1, and Myt1. Second, one or more of these factors must act either directly or indirectly to repress Math5’s expression, thereby establishing a negative feedback loop. Third, the early expressing factors must maintain their

expression in the absence of Math5 by autoand crossregulation, thereby establishing a positive feedback loop.

At present, it is not known whether Math5 is a direct activator of any of the early-expressing transcription factors. In Xenopus, a Pou4f2 orthologue is thought to be a direct target of Xenopus Ath5 (Hutcheson and Vetter, 2001), but no evidence has emerged to indicate that Pou4f2 is a target of Math5 in the mouse retina. Unfortunately, very little is known about the mechanisms that turn off Math5 in RGCs or that maintain the expression of Pou4f2, Isl1, Tle1, and Myt1 following downregulation of Math5 expression. It is possible that Math5 activates the expression of a transcription factor that is positioned genetically upstream of Pou4f2 and the other early-expressing transcription factors, but to date, no compelling evidence exists for this putative factor. Clearly, much additional work is required before we fully understand the regulatory events that advance a competent RPC to a committed RGC.

Genes Regulated by Pou4f2 The best-studied transcription factor associated with the differentiation of RGCs is Pou4f2, which is essential for RGC differentiation, axon outgrowth and pathfinding, and RGC survival (Erkman et al., 1996, 2000; Gan et al., 1996, 1999; Wang et al., 2000, 2002). Although Pou4f2 clearly plays a crucial role in RGC differentiation, RGCs still form in its absence, but they are abnormal and most undergo apoptosis by E18.5. In addition, the expression of many RGC genes is not affected by the absence of Pou4f2 (Mu et al., 2004). These results indicate that parallel regulatory pathways independent of Pou4f2 contribute to RGC differentiation. These other pathways are likely to involve Isl1, Tle1, and Myt1, each representing a different branch of the RGC gene regulatory network that is controlled ultimately by Math5 (see figure 26.2).

Pou4f2 functions as a transcriptional activator (Trieu et al., 1999; Martin et al., 2005), and gene expression profiling with Pou4f2 null retinas has identified 87 genes whose expression is significantly altered in the absence of Pou4f2 (Mu et al., 2004). The encoded products of the Pou4f2-dependent genes fall largely into four major functional groups: neural integrity and function, secreted signaling molecules, transcription factors, and cell cycle regulators. As expected, Pou4f2 is required for the expression of many genes in RGCs, but unexpectedly, Pou4f2 is also required for genes that are expressed in proliferating RPCs where Pou4f2 is never expressed. The dependence of a number of RPC genes on Pou4f2 suggests that Pou4f2 has non-cell-autono- mous functions. These functions are crucial for maintaining the correct number of RPCs by controlling their rate of proliferation (Mu et al., 2004; Mu et al., 2005b).

Constructing a comprehensive RGC gene regulatory network requires distinguishing genes whose expression depends directly on Pou4f2 from those whose dependence