AKE PCA ETR:PERSPECTIVE FEATURE: SPECIAL SACKLER

Regulatory logic of neuronal diversity: Terminal selector and selector motifs

Oliver Hobert1 Howard Hughes Medical Institute, Department of Biochemistry and Molecular Biophysics, Columbia University Medical Center, New York, NY 10032

Edited by Michael S. Levine, University of California, Berkeley, CA, and approved September 8, 2008 (received for review June 24, 2008)

Individual neuronal cell types are defined by the expression of unique batteries of terminal differentiation genes. The elucidation of the cis-regulatory architecture of several distinct, single neuron type-specific batteries in Caenorhabditis elegans has revealed a strikingly simple cis-regulatory logic, in which small cis-regulatory motifs are activated in postmitotic neurons by autoregulating transcription factors (TFs). Loss of the TFs results in the loss of the identity of the individual neuron type. I propose to term these TFs ‘‘terminal selector genes’’ and their cognate cis-regulatory target sites ‘‘terminal selector motifs.’’ Terminal selector genes assign individual neuronal identities by directly controlling the expression of downstream, terminal differentiation genes and act in specific regulatory network configurations. The simplicity of the cis-regulatory logic on which the terminal selector gene concept is based may contribute to the evolvability of neuronal diversity.

Caenorhabditis elegans ͉ neuronal differentiation ͉ gene regulation ͉ transcription factors ͉ cis-regulatory motifs

pproximately a century ago, ‘‘neuron type,’’ which is more frequently AIY Interneurons Ramon y Cajal (1) discovered used in vertebrates) (2). Gene expres- The identity of the AIY class of inter- and described in detail the an- sion data exist for many individual neu- neurons, which is composed of two atomical complexity of cell ron types, and cis-regulatory control bilaterally symmetric interneurons in- Atypes in nervous systems. His work regions can be easily dissected through volved in processing sensory information framed the fundamental question of mutational analysis of reporter genes and in behavioral plasticity (7, 8), is how such cellular complexity is gener- expressed in transgenic animals. controlled by a similarly simple regula- ated during development. The molecular I will first briefly describe several ex- tory logic. A broad battery of genes that correlates to the anatomical diversity of amples in which the regulatory program defines AIY identity shares a small cis- neuron types in a mature nervous sys- of individual neuron types has been elu- regulatory motif, called the AIY motif tem are neuron-type specific gene bat- cidated in C. elegans in substantial detail (Table 1), which, like the ASE motif, is teries. The composition of neuron (Table 1) (3) and will then discuss com- required and sufficient to instruct gene type-specific gene batteries is highly mon principles that may apply across combinatorial. That is, the differentiated phylogeny. expression in AIY (9). Like the ASE properties of individual neuron types motif, the AIY motif commonly occurs are usually not defined by the unique ASE Gustatory Neurons in single copy in the vicinity of AIY- expression of specific gene products, but The ASE class of sensory neurons is expressed genes. The AIY motif is syn- rather by the unique combination of composed of two bilaterally symmetric ergistically activated by a dimer of two genes that may each be more broadly sensory neurons. Their transcriptome homeodomain , the TTX-3 LIM expressed. The question of how neuro- was determined by SAGE analysis (4). homeodomain and the CEH-10 nal diversity is generated, that is, how The analysis of the regulatory regions of Paired-type homeodomain protein (9). individual neurons execute distinct and ASE-expressed genes, performed by re- Each of these factors is expressed in a unique differentiation programs, can porter gene analysis in transgenic few neurons, but their expression exclu- therefore be essentially boiled down to worms, revealed a strikingly simple cis- sively overlaps in the AIY neuron class, the question of how such combinatorial regulatory logic. ASE-expressed genes thereby making them, as is the case for mechanisms are en- contain in their proximity a small cis- CEH-1, a unique identifier for AIY coded in the genome, on both the level regulatory motif of Ϸ12 bp, called the identity (10). In either ttx-3 or ceh-10 of cis-regulatory control elements and ASE motif (Table 1), which is absolutely null mutants, the identity of the mature the trans-acting factors that read these required as well as sufficient by itself to AIY interneuron is completely lost. As control elements. drive gene expression in the ASE neu- in the case of che-1 and the ASE neuron One strategy to tackle this problem is rons (4). This motif is bound and acti- class, the AIY interneurons are gener- to use a bottom-up approach in which vated by a transcription ated and still express pan-neuronal fea- one first defines the nuts-and-bolts gene factor (TF), CHE-1, that is exclusively batteries that define the specific ana- expressed in the ASE sensory neurons. tomical and functional properties of a In the absence of the che-1 gene, ASE This paper results from the Arthur M. Sackler Colloquium of neuron and then dissects the cis- motif-containing genes fail to be acti- the National Academy of Sciences, ‘‘Gene Networks in An- imal Development and Evolution,’’ held February 15–16, regulatory control elements of these vated, resulting in a complete loss of the 2008, at the Arnold and Mabel Beckman Center of the genes. Such an approach is particularly identity of the mature ASE sensory neu- National Academies of Sciences and Engineering in Irvine, feasible in the experimentally easily ron class (4–6). Panneuronal features CA. The complete program and audio files of most presen- amenable and neuroanatomically well are unaffected and no specific alterna- tations are available on the NAS web site at http:// characterized model organism Caeno- tive fate appears to be executed upon www.nasonline.org/SACKLER_Gene_Networks. rhabditis elegans, which contains a ner- loss of che-1, indicating that ASE is Author contributions: O.H. wrote the paper. vous system of 302 neurons that fall into stuck in an indeterminate ground neuro- The author declares no conflict of interest. 118 anatomically precisely defined neu- nal state. Moreover, CHE-1 is sufficient This article is a PNAS Direct Submission. ron classes (the term ‘‘neuron class’’ is to induce ASE neuron fate if expressed 1E-mail: [email protected]. used here interchangeably with the term in other sensory neurons (6). © 2008 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0806070105 PNAS ͉ December 23, 2008 ͉ vol. 105 ͉ no. 51 ͉ 20067–20071 Downloaded by guest on September 29, 2021 Table 1. Examples of neuronal terminal selector genes in C. elegans Sufficiency Terminal selector gene Neuron class Selector motif of motif

CHE-1 zinc finger transcription ASE sensory neurons ASE motif Yes factor

TTX-3/CEH-10 LIM/Prd AIY interneurons AIY motif Yes homeodomain dimer

AST-1 ETS-type transcription All dopaminergic neurons DA motif Yes factor

MEC-3/UNC-86 LIM/POU Mechanosensory neurons Yes homeodomain dimer

UNC-30 Prd-type GABAergic ventral cord ND homeodomain motorneurons

See text for references. ND, not determined.

tures in ttx-3 or ceh-10 mutants and have (Table 1) (17, 18). The identity of all ago, which describes genes that are re- therefore ‘‘only’’ lost their neuron type- dopaminergic neurons in the worm are quired to determine the identity of a specific identity. Ectopic coexpression of specified by an ETS domain TF that specific developmental field or organ. ttx-3 and ceh-10 can transform neurons acts through a defined cis-regulatory The original selector gene definition into an AIY-like fate, but, notably, only motif, the DA motif, that is a required included genes that turned out to act in specific cellular contexts (10). and sufficient determinant for gene ex- relatively early in developmental path- pression in all DA neuron types (N. ways (21), and the definition included Touch Sensory Neurons Flames and O.H., unpublished data). A no specific mechanism of action. By ter- In the most classic example of regula- more detailed phenotypic characteriza- minal selector gene I mean to indicate tory logic of neuronal specification in C. tion of animals carrying mutations in that the gene is not only required to elegans, the identity of a group of six other neuron type-specific TFs, e.g., determine the identity of a specific neu- mechanosensory neurons is determined ttx-1 (AFD neurons), odr-7 (AWA neu- ron type but that it does so by directly by the combinatorial activity of two ho- rons), lim-4 (AWB neurons), and ceh-36 regulating the expression of terminal meobox genes mec-3 and unc-86 (11– (AWC neurons) (3, 19), and the identifi- differentiation genes [such terminal dif- 13). A heterodimer of MEC-3 and cation of their direct target gene batter- ferentiation genes were coined ‘‘realiza- UNC-86 acts synergistically to directly ies may reveal that those factors also tor’’ genes by Garcia-Bellido (20)]. A control the expression of mechanosen- broadly control neuron type-specific terminal differentiation gene battery sory-specific, terminal differentiation gene batteries through simple cis- defines the stable, unique properties of genes through a conserved cis- regulatory motifs. a specific, postmitotic neuron type, is regulatory motif that is required and maintained throughout the life of a neu- sufficient for touch neuron expression Terminal Selector Genes and Terminal ron, and encompasses genes such as (Table 1) (14, 15). This cis-regulatory Selector Motifs neurotransmitter receptors, ion chan- motif was subsequently found to be The examples described above and listed nels, neurotransmitter synthesis pathway present in many other touch-neuron in Table 1 reveal a common underlying genes, structural protein, etc. Loss of expressed genes as well (16). theme of neuronal specification in C. the terminal selector gene results in a elegans. Neuron-type specific gene bat- loss of the specific identity of the neu- Other Examples teries contain a common ‘‘tag’’ in the ron type, without losing overall neuronal The simple regulatory logic described form of a specific cis-regulatory motif, identity, though, as pan-neuronal fea- above may apply broadly throughout the and this motif is controlled by a TF (ei- tures still remain expressed in terminal C. elegans nervous system. GABAergic ther a single factor or a combinatorially selector gene mutants. Gross morphol- motor neurons in the ventral nerve cord acting TF complex) that specifies the ogy and position may also be unaffected are specified by a Prd-type homeo- identity of the neuron. I propose to by terminal selector genes (6, 22), un- domain protein, UNC-30, which also term these TFs ‘‘terminal selector derscoring their role in determining ter- appears to directly control the expres- genes.’’ This term represents an exten- minal identities rather than properties sion of terminal differentiation markers sion of the selector gene concept intro- that are already predetermined in pro- through a small cis-regulatory motif duced by Garcia-Bellido (20) Ͼ30 years genitor stages. Neuronal terminal selec-

20068 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0806070105 Hobert Downloaded by guest on September 29, 2021 fied in vertebrates by using transcrip- tome analysis from isolated cells (30). However, cis-regulatory control regions of such terminal differentiation gene batteries have not been analyzed on a scale that is required to draw conclu- sions about common regulatory features, such as the presence of terminal selector motifs. A striking exception is the verte- brate retina, which is composed of well characterized individual neuron types (31). Combinations of transcriptome analysis in wild-type and mutant back- grounds and bioinformatic analysis have revealed a terminal selector gene that appears to be very akin to the inverte- brate examples discussed above, the Crx Paired-type homeodomain protein (32– 34). Crx is a terminal selector gene that Fig. 1. Network configuration of terminal selector genes. Terminal selector TFs (acting either alone or controls photoreceptor fate by directly in synergistic combination) activate downstream target genes directly via terminal selector motifs and also regulating scores of terminal differentia- autoregulate their own expression via those motifs. Autoregulated, maintained expression of a terminal selector is critical to maintain the differentiated features of the cell. Downstream targets of terminal tion genes, acting also through a small selectors (X) define differentiated properties of a neuron, such as neurotransmitter , ion channels, cis-regulatory motif. Crx may cooperate adhesion proteins etc. Targets may also include TFs that regulate specific ‘‘subroutines.’’ The targets of in its terminal selector gene function these TFs may not harbor terminal selector motifs; therefore, not all genes expressed in a given neuron with the closely related Prd- type must have a selector motif in their regulatory region (4). TFs that are induced by terminal selectors gene Otx2 (35, 36). The Crx/Otx2 termi- may also cooperate with terminal selector proteins in a feed-forward loop configuration to jointly control nal selector gene function appears to be specific terminal genes (39, 44). Besides activating the expression of identity-defining genes it is also conserved in flies as well (37, 38). The possible that terminal selectors may repress alternative fates through repressing the expression of selector mouse homeobox genes Tlx3 and Tlx1, genes for other cell types. This would explain neuronal identity switches observed upon removing putative which determine postmitotic neuronal terminal selector genes (26). identity in the spinal cord (26), are among the many other known vertebrate tor genes act through what I propose to type. The mechanistic basis for the lat- TFs that are candidates to classify as term ‘‘terminal selector motifs,’’ the ter scenario may be that terminal selec- terminal selector genes as well; identifi- simple cis-regulatory motifs described tor genes may not only activate terminal cation of their direct target genes and a above (Table 1). Terminal selector differentiation genes, but also inhibit more extensive mutant phenotypic char- genes are expressed throughout the life alternative fates by repressing the ex- acterization will demonstrate whether of the neuron and are continuously re- pression of other terminal selector these genes indeed fall into this quired for maintaining the differentiated genes. For example, vertebrate Tlx3 and category. state of a neuron. This maintenance Tlx1 genes may act in this manner (26). Integration of Terminal Selector Genes function is mediated by autoregulation The broad effect that terminal selec- into Regulatory Networks through terminal selector motifs that are tor TFs have on the identity of a neuron Gene regulatory networks have increas- present in the promoter of the terminal sets them apart from the plethora of ingly become recognized as an assembly selector genes themselves (4, 9, 23) TFs that directly control specific subrou- of smaller network motifs, which endow (Fig. 1). tines of a neuron type and may them- these systems with specific functional The terminal selector terminology selves be targets of terminal selector properties (39, 40). Terminal selector only encompasses genes that directly TFs. For example, the ceh-23 homeodo- gene function is characterized by the bind to the regulatory regions of postmi- main protein is a direct target of the occurrence of three different network totically expressed, terminal differentia- TTX-3/CEH-10 terminal selector com- motifs (Fig. 1): (i) positive autoregula- tion batteries, thereby setting it apart plex in the C. elegans AIY neuron class. tory motifs, i.e., terminal selector genes from the ‘‘neuron-type selector genes’’ Loss of ceh-23 affects the expression of regulate their own expression (4, 22, coined by Jan and Jan (24) based on a specific transmembrane receptor re- 23); (ii) single input modules (SIMs), work in Drosophila or from the many quired for neuronal plasticity but does which indicate the coregulation of the TFs, in both invertebrates and verte- not affect any other known differenti- terminal gene battery; and (iii) feed- brates, that act at various upstream ated property of AIY (8, 10). forward loop motifs in which a terminal steps in neuronal commitment and de- Do neuronal terminal selector gene selector gene controls the expression of termination and define neuronal pro- exist in more complex nervous systems another TF and then cooperates with genitor identities (25). as well? The selector gene concept has this TF to regulate terminal differentia- The loss of a terminal selector gene already been applied to TFs acting in tion genes. Examples include the above- results in a broad identity loss, as envi- neuronal specification in flies and verte- mentioned vertebrate Crx TF that sioned in the original Garcia-Bellido brates (24, 26–29), yet it is not known controls expression of the Nrl TF, which selector concept. Identity loss entails whether in those cases the respective TF jointly control the expression of target either the conversion into a nonrecog- is truly a terminal selector gene that genes in rod photoreceptors (32). Other nizable undifferentiated state, as ap- directly acts on terminal differentiation examples are the C. elegans terminal pears to be the case for the examples gene batteries. Terminal differentiation selector gene complex TTX-3/CEH-10, listed in Table 1, or it may entail the gene batteries that define individual which regulates the homeodomain pro- switch to the identity of another neuron neuron types are beginning to be identi- tein CEH-23 to then jointly regulate the

Hobert PNAS ͉ December 23, 2008 ͉ vol. 105 ͉ no. 51 ͉ 20069 Downloaded by guest on September 29, 2021 sra-11 downstream target gene (10) or which globally controls sensory neuron- control regions, upstream inputs into C. the CHE-1 terminal selector that di- wide subroutines, such as the expression elegans terminal selectors are complex rectly controls expression of the CEH-36 of genes that build sensory cilia (45). and composed of multiple different cis- Otx-type homeodomain protein to then Pan-neuronal features may be controlled regulatory elements that may serve to jointly regulate with CEH-36 the expres- by a common trans-acting factor as sug- sample the developmental history of the sion of the target gene gcy-7. gested by the cis-regulatory analysis of neuron (48, 49) (unpublished data). The Feed-forward motifs and the implicit some pan-neuronal genes (46), but oth- combination of multiple outputs and existence of additional layers of regula- ers may be controlled in a more piece- multiple inputs results in an hourglass- tory control downstream of terminal meal manner by as-yet-unknown factors shaped regulatory topology, which can selector genes define another property (47). The existence of parallel regula- be considered to be one of the defining of terminal selector genes. Neuron tory programs illustrates that terminal feature of terminal selector genes classes, whose identity is controlled by selector genes act by determining the (Fig. 1). specific terminal selector genes, can of- specific identity of neuron types, rather Terminal selector genes can also have ten be subdivided into specific subclass- than by determining broad and generic functions upstream of their terminal dif- es/subtypes. These subclasses essentially features of all or large groups of ferentiation function and thereby act represent a variation of a theme, as il- neurons. repeatedly in a given lineage. For exam- lustrated by the example of vertebrate ple, the unc-86 terminal selector gene, a photoreceptors and the ASE gustatory The Importance of Context. The activity of POU homeobox gene, acts early in spe- neurons in C. elegans. Vertebrate photo- neuronal terminal selector genes is modu- cific neuronal lineages to determine the receptors, whose identity is controlled lated by cellular context. Terminal selec- overall identity of the several neuro- by the terminal selector Crx, can be sub- tor genes, such as those shown in Table 1, blasts (13, 14). Within a subset of those divided into two subclasses, cone and can act only in specific cellular contexts lineages, UNC-86 then acts several divi- rod cells. The identity of the rod sub- and apparently only at specific develop- sions later, now with another homeo- class is controlled by a direct target of mental time points, as revealed, for exam- domain protein, MEC-3, as a terminal Crx, the Nrl basic protein ple, by misexpression studies (9, 14). selector complex to control the terminal (41). Crx and Nrl then cooperate in a Moreover, genes may act as terminal se- specification of one specific neuron feed-forward configuration to regulate lector genes in one cell type, but may type, touch sensory neurons (13, 14). rod-specific genes (32). Nrl also up- have no such function in other neuron This example also illustrates again the regulates NR2E2, a TF types in which they are expressed. For context dependency of terminal selector that represses a subset of cone genes in example, the ast-1 gene is a terminal se- genes; in the context of several neuro- rods (42, 43). Another example is the C. lector gene for dopaminergic neuron fate, blasts, unc-86 acts to control neuroblast elegans ASE gustatory neuron class that but in some nondopaminergic neurons it identity (likely in conjunction with as- can be subdivided into two distinct sub- may have much more restricted functions yet-unknown TFs), whereas terminally classes, the left ASE neuron (ASEL) in controlling axon outgrowth, but not differentiating touch neurons provide and the right ASE (ASER) neuron, overall cell fate (N. Flames and O.H., un- the context in the form of the MEC-3 which are distinguishable by the expres- published data). Likely context determi- homeodomain protein with which sion of a different set of putative nants are the presence of other regulatory UNC-86 heterodimerizes to control ter- chemoreceptors (44). The C. elegans factors with which terminal selector TFs minal touch neuron differentiation. terminal selector gene CHE-1 controls may directly interact. For example, the overall ASE fate, but also directly con- ttx-3 homeobox selector gene is expressed Evolvability trols the expression of a set of regula- in five neuron classes, but acts as a selec- Gene regulation by terminal selector tory factors that distinguish ASEL from tor gene in only the one neuron class in genes provides a framework to think ASER and then cooperates with these which its interaction partner, the ceh-10 about the evolution of neuronal diversity. factors to directly regulate downstream homeobox gene, is present (10). Context The simplicity of terminal selector motifs target genes specific for ASEL or dependency is also evident from the per- may mean that individual genes can be ASER (44). spective of selector motifs. In all cases rapidly recruited into or lost from a neu- Taken together, terminal selector tested selector motifs are alone sufficient ron type-specific gene battery through loss genes directly control the expression of to instruct neuron type-specific expression or gain of terminal selector motifs. Such the core set of identity-defining nuts- if placed into the context of a reporter plasticity may provide a rich playground and-bolts terminal differentiation (real- gene vector (Table 1) (4, 9, 14) (N. for evolution. The small size of terminal izator) genes, but they also control Flames and O.H., unpublished data). selector motifs and their consequent downstream regulatory events that fur- However, the small size of terminal selec- abundance in genomes suggests that many ther diversify individual neuron types. tor motifs and henceforth their abundance terminal selector motifs are not active; in genomic sequences (4) demands the whatever regulates this dormancy, such as Parallel Regulatory Programs existence of an additional layer of reg- the above-mentioned possible nucleo- The genetic removal of neuronal termi- ulatory control, of which nucleosome- some-mediated motif accessibility, nal selector genes reveals the existence dependent binding accessibility is one provides another regulatory layer for evo- of regulatory programs that act in paral- possibility. lution to play on. The addition of a func- lel to terminal selector genes and are tional selector motif into the control therefore unaffected by the loss of the Upstream of Terminal Selectors region of a TF may also add a specific terminal selector. For example, neurons Many layers of transcriptional control ‘‘subroutine’’ to an existing neuronal class- lacking terminal selector genes shown in mechanism funnel into the determina- specific gene expression program. An- Table 1 still express panneuronal fea- tion of a neuron type (25), and I pro- other way to view terminal selector genes tures and, in the case of sensory neuron pose here that terminal selector genes in an evolutionary context is that the re- terminal selectors, may still express pan- represent the regulatory endpoint of cruitment of terminal selector genes into a sensory features (4, 9, 14). The latter these complex gene regulatory cascades novel cellular context, e.g., by the acquisi- features are controlled by the phyloge- (Fig. 1). As revealed by mutant analysis tion of a novel cis-regulatory input into netically conserved RFX-type TF daf-19, and the dissection of their cis-regulatory the terminal selector gene locus may gen-

20070 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0806070105 Hobert Downloaded by guest on September 29, 2021 erate a dramatically, rather than incre- into the terminal selector gene concept. and cognate terminal selector genes in mentally distinct, novel neuronal cell type, More neuron types need to be examined other regions of the vertebrate brain. As defined by the set of genes originally ex- as it is well conceivable that cases may the terminal selector function of a TF pressed in the cell type plus an entirely surface in which neuronal identity is may be obscured by earlier functions of novel gene battery; such a ‘‘hybrid’’ cell is determined in a more ‘‘piecemeal’’ man- the gene, temporally controlled gene knockouts are required to put the termi- then subjected to Darwinian selection. ner. In contrast to C. elegans, the group- nal selector gene concept to a rigorous Taken together, simplicity in regulatory ing of neurons into specific classes is a test in vertebrates. Finally, even though architecture may provide evolutionary much harder task in vertebrates. In the I have discussed the terminal selector plasticity. most extensively charted vertebrate neu- gene concept in the context only of the ronal domain, the retina, neuron types nervous system, it may apply to diversify Outlook can be relatively clearly defined (31) and other cellular fates as well. For example, The morphologically and lineally com- evidence for terminal selector genes in- the homeodomain TF Pax5 may serve as pletely mapped out nervous system of C. deed appears to exist, as discussed a terminal selector gene for B cell fate elegans and its amenability to genetic above. The identification and functional determination in the immune system (50). analysis facilitated the identification of analysis of vertebrate TFs that fulfill the ACKNOWLEDGMENTS. I thank R. Mann, P. Sen- terminal selector genes in C. elegans. most fundamental criterion for a termi- gupta, C. Cepko, C. Desplan, I. Greenwald, and Y. Most neuron classes studied in depth in nal selector gene, i.e., the maintained Jin for comments on the manuscript. My work is supported by the Howard Hughes Medical Insti- C. elegans have revealed a common reg- expression throughout the life of a neu- tute and National Institutes of Health Grants ulatory principle that can be subsumed ron, may aid in defining neuron classes R01NS039996-05 and R01NS050266-03.

1. Cajal Ry (1911) Histologie du Syste`me Nerveux de 18. Cinar H, Keles S, Jin Y (2005) Expression profiling of 35. Nishida A, et al. (2003) Otx2 homeobox gene controls L’Homme et des Verte´bre´s (Maloine, Paris). GABAergic motor neurons in Caenorhabditis elegans. retinal photoreceptor cell fate and pineal gland devel- 2. White JG, Southgate E, Thomson JN, Brenner S (1986) Curr Biol 15:340–346. opment. Nat Neurosci 6:1255–1263. The structure of the nervous system of the nematode 19. Lanjuin A, Sengupta P (2004) Specification of chemo- 36. Koike C, et al. (2007) Functional roles of Otx2 transcrip- Caenorhabditis elegans. Philos Trans R Soc London Ser sensory neuron subtype identities in Caenorhabditis tion factor in postnatal mouse retinal development. B 314:1–340. elegans. Curr Opin Neurobiol 14:22–30. Mol Cell Biol 27:8318–8329. 3. Hobert O (2005) Specification of the nervous system. 20. Garcia-Bellido A (1975) Genetic control of wing disc de- 37. Ranade SS, et al. (2008) Analysis of the Otd-dependent WormBook, eds The C. elegans Research Community, velopment in Drosophila. Ciba Found Symp 29:161–182. transcriptome supports the evolutionary conservation doi/10.1895/wormbook.1.12.1. 21. Mann RS, Carroll SB (2002) Molecular mechanisms of of CRX/OTX/OTD functions in flies and vertebrates. Dev 4. Etchberger JF, et al. (2007) The molecular signature and selector gene function and evolution. Curr Opin Genet Biol 315:521–534. cis-regulatory architecture of a C. elegans gustatory Dev 12:592–600. 38. Tahayato A, et al. (2003) Otd/Crx, a dual regulator for neuron. Genes Dev 21:1653–1674. 22. Hobert O, et al. (1997) Regulation of interneuron func- the specification of ommatidia subtypes in the Dro- 5. Chang S, Johnston RJ, Jr, Hobert O (2003) A transcrip- tion in the C. elegans thermoregulatory pathway by sophila retina. Dev Cell 5:391–402. tional regulatory cascade that controls left/right asym- the ttx-3 LIM homeobox gene. Neuron 19:345–357. 39. Alon U (2007) Network motifs: Theory and experimen- metry in chemosensory neurons of C. elegans. Genes 23. Way JC, Chalfie M (1989) The mec-3 gene of Caeno- tal approaches. Nat Rev Genet 8:450–461. Dev 17:2123–2137. rhabditis elegans requires its own product for main- 40. Davidson EH (2001) Genomic Regulatory Systems (Ac- 6. Uchida O, Nakano H, Koga M, Ohshima Y (2003) The C. tained expression and is expressed in three neuronal ademic, San Diego). elegans che-1 gene encodes a zinc finger required for specification of the ASE chemosen- cell types. Genes Dev 3:1823–1833. 41. Mears AJ, et al. (2001) Nrl is required for rod photore- sory neurons. Development 130:1215–1224. 24. Jan YN, Jan LY (1994) Neuronal cell fate specification in ceptor development. Nat Genet 29:447–452. 7. Tsalik EL, Hobert O (2003) Functional mapping of neu- Drosophila. Curr Opin Neurobiol 4:8–13. 42. Corbo JC, Cepko CL (2005) A hybrid photoreceptor rons that control locomotory behavior in Caenorhab- 25. Guillemot F (2007) Spatial and temporal specification expressing both rod and cone genes in a mouse model ditis elegans. J Neurobiol 56:178–197. of neural fates by transcription factor codes. Develop- of enhanced S-cone syndrome. PLoS Genet 1:e11. 8. Remy JJ, Hobert O (2005) An interneuronal chemore- ment 134:3771–3780. 43. Chen J, Rattner A, Nathans J (2005) The rod photore- ceptor required for olfactory imprinting in C. elegans. 26. Cheng L, et al. (2004) Tlx3 and Tlx1 are postmitotic ceptor-specific nuclear receptor Nr2e3 represses tran- Science 309:787–790. selector genes determining glutamatergic over scription of multiple cone-specific genes. J Neurosci 9. Wenick AS, Hobert O (2004) Genomic cis-regulatory GABAergic cell fates. Nat Neurosci 7:510–517. 25:118–129. architecture and trans-acting regulators of a single 27. Mangale VS, et al. (2008) Lhx2 selector activity specifies 44. Hobert O (2006) Architecture of a microRNA- interneuron-specific gene battery in C. elegans. Dev cortical identity and suppresses hippocampal organizer controlled gene regulatory network that diversifies Cell 6:757–770. fate. Science 319:304–309. neuronal cell fates. Cold Spring Harb Symp Quant 10. Altun-Gultekin Z, et al. (2001) A regulatory cascade of 28. Sieber MA, et al. (2007) Lbx1 acts as a selector gene in Biol 71:181–188. three homeobox genes, ceh-10, ttx-3, and ceh-23, con- the fate determination of somatosensory and visceros- 45. Swoboda P, Adler HT, Thomas JH (2000) The RFX-type trols cell fate specification of a defined interneuron ensory relay neurons in the hindbrain. J Neurosci transcription factor DAF-19 regulates sensory neuron class in C. elegans. Development 128:1951–1969. 27:4902–4909. cilium formation in C. elegans. Mol Cell 5:411–421. 11. Way JC, Chalfie M (1988) mec-3, a homeobox-contain- 29. Nakatani T, Minaki Y, Kumai M, Ono Y (2007) Helt 46. Ruvinsky I, Ohler U, Burge CB, Ruvkun G (2007) Detec- ing gene that specifies differentiation of the touch determines GABAergic over glutamatergic neuronal tion of broadly expressed neuronal genes in C. elegans. receptor neurons in C. elegans. Cell 54:5–16. fate by repressing Ngn genes in the developing mes- Dev Biol 302:617–626. 12. Chalfie M, Horvitz HR, Sulston JE (1981) Mutations that encephalon. Development 134:2783–2793. 47. Hwang SB, Lee J (2003) Neuron cell type-specific lead to reiterations in the cell lineages of C. elegans. 30. Nelson SB, Hempel C, Sugino K (2006) Probing the SNAP-25 expression driven by multiple regulatory ele- Cell 24:59–69. transcriptome of neuronal cell types. Curr Opin Neu- ments in the nematode Caenorhabditis elegans. J Mol 13. Finney M, Ruvkun G (1990) The unc-86 gene product robiol 16:571–576. Biol 333:237–247. couples cell lineage and cell identity in C. elegans. Cell 31. Masland RH (2004) Neuronal cell types. Curr Biol 48. Mitani S, Du H, Hall DH, Driscoll M, Chalfie M (1993) 63:895–905. 14:R497–R500. Combinatorial control of touch receptor neuron ex- 14. Duggan A, Ma C, Chalfie M (1998) Regulation of touch 32. Hsiau TH, et al. (2007) The cis-regulatory logic of the pression in Caenorhabditis elegans. Development receptor differentiation by the Caenorhabditis elegans mec-3 and unc-86 genes. Development 125:4107–4119. mammalian photoreceptor transcriptional network. 119:773–783. 15. Xue D, Finney M, Ruvkun G, Chalfie M (1992) Regula- PLoS ONE 2:e643. 49. Baumeister R, Liu Y, Ruvkun G (1996) Lineage-specific tion of the mec-3 gene by the C. elegans homeopro- 33. Chen S, et al. (1997) Crx, a novel Otx-like paired- regulators couple cell lineage asymmetry to the tran- teins UNC-86 and MEC-3. EMBO J 11:4969–4979. homeodomain protein, binds to and transactivates pho- scription of the Caenorhabditis elegans POU gene 16. Zhang Y, et al. (2002) Identification of genes expressed in toreceptor cell-specific genes. Neuron 19:1017–1030. unc-86 during neurogenesis. Genes Dev 10:1395– C. elegans touch receptor neurons. Nature 418:331–335. 34. Blackshaw S, Fraioli RE, Furukawa T, Cepko CL (2001) 1410. 17. Jin Y, Hoskins R, Horvitz HR (1994) Control of type-D Comprehensive analysis of photoreceptor gene expres- 50. Hagman J, Lukin K (2007) ‘‘Hands-on’’ regulation of B GABAergic neuron differentiation by C. elegans sion and the identification of candidate retinal disease cell development by the transcription factor Pax5. Im- UNC-30 homeodomain protein. Nature 372:780–783. genes. Cell 107:579–589. munity 27:8–10.

Hobert PNAS ͉ December 23, 2008 ͉ vol. 105 ͉ no. 51 ͉ 20071 Downloaded by guest on September 29, 2021