Specification of Neurotransmitter Identity by Phox2 Proteins in Neural Crest Stem Cells

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Neuron, Vol. 22, 693–705, April, 1999, Copyright 1999 by Cell Press Specification of Neurotransmitter Identity by Phox2 Proteins in Neural Crest Stem Cells

Liching Lo,* Xavier Morin,† Jean-Franc¸ ois Brunet,† others (Sharma et al., 1998). It is also not clear how and David J. Anderson*‡ the specification of such subtype-specific properties is *Division of Biology 216–76 linked to the programs that control generic neuronal Howard Hughes Medical Institute differentiation. California Institute of Technology A deeper understanding of the regulatory circuits that Pasadena, California 91125 specify neuronal identity is limited by the lack of infor- † Laboratorie de Ge´ ne´ tique et Physiologie mation about direct interactions between transcription du De´ veloppement factors and terminal differentiation genes expressed in Institut de Biologie du De´ veloppement subsets of neurons. Relatively few direct targets of ge- de Marseille netically defined transcriptional regulators of neuronal Centre National de la Recherche identity have been identified (see, e.g., Mitani et al., Scientifique/Institut National de la Sante´ 1993). Conversely, few cis-acting regulatory elements et de la Recherche Me´ dicale/Universite´ de governing neuronal subtype-specific expression of ter- la Me´ diterrane´ e minal differentiation genes have been characterized (re- AP de Marseille viewed by Mandel and McKinnon, 1993; Quinn, 1996). Marseille It is therefore not clear, for example, whether different France classes of transcription factors directly regulate the expression of either subtype-specific or panneuronal properties or whether the same factor(s) regulates both. Summary We have used the specification of sympathetic autonomic neuron identity as a model system for investigat- We have investigated the specification of noradrener- ing the mechanisms that determine different compo- gic neurotransmitter identity in neural crest stem cells nents of neuronal phenotype. Previously, we showed (NCSCs). Retroviral expression of both wild-type and that an extracellular signal, bone morphogenetic protein dominant-negative forms of the paired homeodomain 2 (BMP2), and two transcription factors, MASH1 (a bHLH transcription factor Phox2a indicates a crucial and di- protein [Johnson et al., 1990]) and Phox2a (a paired rect role for this protein (and/or the closely related homeodomain protein [Valarche´ et al., 1993]), regulate Phox2b) in the regulation of endogenous tyrosine hy- some aspects of autonomic identity. In cultures of iso- lated neural crest stem cells (NCSCs; Stemple and An- droxylase (TH) and dopamine-␤ hydroxylase (DBH) gene expression in these cells. In collaboration with derson, 1992), BMP2 induces MASH1 (Shah et al., 1996), cAMP, Phox2a can induce expression of TH but not which in turn activates Phox2a (Lo et al., 1998). This of DBH or of panneuronal genes. Phox2 proteins are, cascade leads to expression of both panneuronal and moreover, necessary for the induction of both TH and some autonomic subtype-specific genes. MASH1 ap- DBH by bone morphogenetic protein 2 (BMP2) (which pears sufficient to activate both classes of genes, while induces Phox2a/b) and forskolin. They are also neces- Phox2a activates only the latter (Lo et al., 1998). sary for neuronal differentiation. These data suggest These studies left open the question of the relation- that Phox2a/b coordinates the specification of neuro- ship of these regulatory molecules to the specification transmitter identity and neuronal fate by cooperating of noradrenergic neurotransmitter identity. Sympathetic environmental signals in sympathetic neuroblasts. neurons express two related genes, tyrosine hydroxylase (TH) and dopamine-␤ hydroxylase (DBH), which control noradrenaline synthesis in these cells. In other Introduction studies, BMP2, MASH1, and Phox2a have each been linked to the regulation of these genes (Guillemot et al., Neuronal identities can be distinguished by a broad 1993; Reissman et al., 1996; Varley and Maxwell, 1996; array of phenotypic properties or components, including Hirsch et al., 1998; reviewed by Goridis and Brunet, morphology, connectivity, neurotransmitter synthesis 1999). Phox2a, in particular, is a leading candidate for and receptivity, and membrane excitability. How these a direct transcriptional activator of TH and DBH. The different components are specified is not yet clear (An- TH and DBH promoters each contain binding sites for derson and Jan, 1997; Edlund and Jessell, 1999). Some Phox2a (also known as Arix [Zellmer et al., 1995]) and transcription factors appear sufficient to specify multi- can be transactivated by this protein in nonneuronal cell ple components of neuronal identity, such as neuro- lines (Zellmer et al., 1995; Kim et al., 1998; Yang et al., transmitter phenotype and axonal trajectory (Thor and 1998). Furthermore, a Phox2a knockout blocks transient Thomas, 1997; Tanabe et al., 1998). Whether such coor- DBH expression in cranial sensory ganglia (Morin et dinate regulation is direct or indirect is not yet clear. al., 1997). Despite this, forced expression of Phox2a is Indeed, certain transcriptional regulators appear to con- insufficient to promote DBH expression in NCSCs (Lo trol some aspects of neuronal identity independent of et al., 1998). Interestingly, it was recently shown that transactivation of the DBH promoter by Phox2a can be ‡ To whom correspondence should be addressed (e-mail: mancusog@ synergistically potentiated by protein kinase A (PKA; cco.caltech.edu). Swanson et al., 1997). These data suggested that cAMP Neuron 694

Figure 1. THϩ Cells in E14.5 Sympathetic Ganglia Coexpress Phox2a and c-RET Freshly dissociated suspensions of cells from thoracic sympathetic ganglia of E14.5 rat fe- tuses were double labeled with anti-TH (A) and anti-Phox2a (B) or anti-TH (C) and anti- c-RET (D). Arrows indicate double-positive cells; arrowheads in (B) indicate Phox2aϩ cells that are TH-. See Table 1 for quantitation. Objective magnification, 40ϫ.

and Phox2a might collaborate to specify noradrenergic 1B, arrows; Table 1). Conversely, 63% of Phox2aϩ cells neurotransmitter identity in NCSCs. coexpressed TH in SG, and in the adrenal glands, the We now show that forced expression of Phox2a in value was 76%. Thus, all THϩ cells are Phox2aϩ, and the presence of forskolin promotes expression of the most Phox2aϩ cells are THϩ. endogenous TH gene in NCSCs but not of DBH. How- We also analyzed expression of TH in relation to ever, BMP2 (which induces Phox2a [Lo et al., 1998] and c-RET, a signal-transducing receptor for glial cell line– 2b) together with forskolin promotes the differentiation derived neurotrophic factor (GDNF; Lindsay and Yanco- of neurons that coexpress both TH and DBH. A domi- poulos, 1996), whose expression is controlled (directly nant-negative form of Phox2a prevents expression of or indirectly) by Phox2a (Morin et al., 1997; Lo et al., both TH and DBH under these conditions and blocks 1998). Virtually all (97%) c-RETϩ cells were THϩ in SG BMP2-induced neuronal differentiation as well. Taken (Figures 1C and 1D, arrows) and adrenal glands (90%; together with previous studies (Zellmer et al., 1995; Table 1), and conversely, all THϩ cells were c-RETϩ in Swanson et al., 1997; Kim et al., 1998; Yang et al., 1998), both tissues (99% and 100%, respectively; Table 1). these data indicate that Phox2 proteins are likely direct Taken together, therefore, these data indicate that most regulators of noradrenergic neurotransmitter identity in or all THϩ cells in SG and adrenal gland coexpress sympathetic neurons and that they are also necessary Phox2a and c-RET. In a previous study, we demon- but not sufficient for generic neuronal differentiation. strated a similarly tight correlation between Phox2a and c-RET expression (Lo et al., 1998). Results Forced Expression of Phox2a Promotes TH TH and Phox2a Are Coexpressed Expression in NCSCs in the Presence in Sympathoadrenal Derivatives of Forskolin of the Neural Crest To determine whether Phox2a was sufficient to promote We first sought to confirm that TH and Phox2a are in fact expression of endogenous TH in neural crest cells, we coexpressed by developing sympathetic neuroblasts. constitutively expressed the transcription factor in these Previously, expression of Phox2a was shown to precede cells using a retroviral vector (Lo et al., 1998; see Experi- and overlap that of TH in sections of developing sympa- mental Procedures). A Myc epitope tag was included in thetic ganglia (Tiveron et al., 1996; Pattyn et al., 1997; the transgene to distinguish its expression from that of %10ف ,Hirsch et al., 1998; Lo et al., 1998). To quantitate the endogenous Phox2a. Under standard conditions coexpression of these markers in individual cells, we of Phox2a-infected clones contained at least some THϩ doubly immunostained acutely dissociated embryonic cells (Figure 2G, “No add”), and about 2.5% of all mycϩ day 14.5 (E14.5) rat sympathetic ganglia (SG) and adre- cells were THϩ. Despite this small effect, no THϩ cells nal glands, fixed a few hours after plating. In SG, 99% were ever observed in uninfected or in control nuclear (129/130) of all THϩ cells were Phox2aϩ, and in adrenal green fluorescent protein– (GFP-) infected clones (Fig- glands, the number was 100% (53/53) (Figures 1A and ure 2G, “GFP”; data not shown). Therefore, Phox2a is Phox2 Proteins Regulate Neurotransmitter Identity 695

Table 1. Coexpression of Phox2a, TH, and c-RET in E14.5 Sympathoadrenal Derivatives SG Adrenal Gland Phox2aϩ THϩ c-RETϩ Phox2aϩ THϩ c-RETϩ TH 99% (129/130) 100% 99% (146/148) 100% (53/53) 100% 100% (103/103) Phox2a 100% 63% (129/206) ND 100% 76% (53/70) ND c-RET ND 97% (146/151) 100% ND 90% (103/114) 100% The values indicate the percentage of cells expressing the markers indicated in the left column that coexpressed the markers indicated at the top. E14.5 lumbar sympathetic ganglia (SG) and adrenal glands were dissociated to single-cell suspensions, plated for about 3 hr, and double labeled with the indicated pairwise combinations of antibodies. Abbreviation: ND, not determined.

sufficient to induce TH expression in at least some A Dominant-Negative Form of Phox2a Interferes NCSCs. with Expression of Both Exogenous Since elevated intracellular cAMP has been shown to and Endogenous TH enhance the ability of Phox2a to transactivate TH re- We wished to determine whether Phox2a is necessary porter constructs (Swanson et al., 1997), we asked for TH expression. Because Phox2b (Pattyn et al., 1997) whether addition of forskolin could potentiate TH induc- might compensate for the absence of Phox2a, we de- tion by Phox2a. Forskolin (5 ␮M) increased to almost signed a dominant-negative form of Phox2a that should 50% the percentage of infected clones containing THϩ interfere with the function of both proteins. To this end, cells (Figures 2E and 2G). Even in the presence of for- a fusion protein was constructed containing the DNA- skolin, however, no induction of TH was seen in unin- binding domain of Phox2a (which is virtually identical fected or in control GFP–infected cells (Figures 2B, 2D, to that of Phox2b) fused to the engrailed repressor (ER) 2F, and 2G, “GFP”). A time course of Phox2a-infected domain (Han and Manley, 1993) and called engrailed cells grown in the presence of forskolin indicated that repressor–Phox (ERPH). Such a fusion should convert the first THϩ cells could be detected as early as 24 hr, Phox2a from a transactivator to a repressor. This ap- and maximal numbers were obtained by 96 hr (Figure proach has been successfully used to generate domi- 2H). Forskolin appeared to increase the rate as well as nant-negative forms of a number of other transcription the extent of TH induction (Figure 2H). Thus, Phox2a is factors (see, e.g., Badiani et al., 1994; Bellefroid et al., sufficient to induce TH expression in a subset of NCSCs, 1996; Furukawa et al., 1997). and this effect can be strongly potentiated by forskolin. We first tested the inhibitory activity of ERPH toward The percentage of Phox2a-infected clones containing Phox2a using a transient cotransfection assay. Trans- THϩ cells could be further increased to 60% (Figure fection of 10T1/2 fibroblasts with a TH-lacZ reporter 2G) by inclusion together with forskolin of GDNF and plasmid (see Experimental Procedures) yielded numer- dexamethasone (Dex), factors that have been shown to ous ␤-galactosidaseϩ cells in the presence but not in enhance TH expression in other cell settings (Smith and the absence of a Phox2a expression construct (data not Fauquet, 1984; Lewis et al., 1987; Maxwell et al., 1996; shown). But in the presence of a 2-fold molar excess of Osaka and Sabban, 1997). These factors had relatively an expression plasmid encoding ERPH, such Phox2a- weak effects on their own (Figure 2G). Their actions driven expression of TH-lacZ was virtually completely appear additive rather than synergistic. Thus, multiple abolished (Table 2). In contrast, such inhibition was not environmental factors can potentiate the induction of TH obtained when a 2-fold excess of expression plasmid expression by Phox2a, but the predominant influence encoding GFP or the ER domain alone was used (Table under these conditions appears to be elevated intracel- 2). Thus, ERPH dominantly inhibited the ability of Phox2a lular cAMP. to activate a TH promoter–reporter construct. One explanation for the effect of forskolin is that the We next sought to test the ability of ERPH to inhibit retroviral Phox2a protein is inactive in the absence of expression of the endogenous TH gene rather than of cAMP. However, we previously showed that forced ex- a transfected reporter construct. Initial attempts to in- pression of Phox2a in NCSCs is sufficient to activate hibit TH expression in cell lines that constitutively ex- expression of c-RET, even in the absence of forskolin press the gene, such as MAH cells (Birren and Anderson, (Lo et al., 1998). It was possible that NCSCs are hetero- 1990), were unsuccessful (data not shown). We consid- geneous and that forskolin permitted Phox2a to activate ered that this failure might reflect an inability of ERPH TH in a different population of cells than expressed to displace Phox2 proteins once assembled into a stable c-RET. However, all TH-expressing infected cells also transcription complex on the TH promoter. We therefore expressed c-RET (Figures 3B and 3C, arrows). Thus, the sought a cell line in which expression of endogenous differential requirement of cAMP for activation of TH TH could be induced de novo, so that expression of and c-RET by Phox2a likely reflects differences in the ERPH could be established before such transcriptional requirements of these genes for cofactors activated by complexes would be assembled. Such a requirement cAMP. In contrast to TH, expression of DBH was never was met by 2D10 cells, a line derived by v-myc-immor- detected in Phox2a-infected NCSCs, even under opti- talization of rat neural crest cells (Lo et al., 1990). Endog- mal conditions (data not shown). Such cells also failed enous TH expression is undetectable by Western blot- to express panneuronal markers such as NF160 and ting or immunocytochemistry in these cells but can be peripherin (Lo et al., 1998; data not shown) and had a induced by fibroblast growth factor (FGF) plus Dex (Lo nonneuronal morphology (Figure 3A). et al., 1990). Neuron 696

Figure 2. Induction of TH Expression in NCSCs by Phox2a in the Presence of Forskolin Representative examples of colonies infected with either the Phox2a (A, C, and E) or control GFP (B, D, and F) retrovirus after growth for 96 hr in the presence of forskolin. (A) and (B), (C) and (D), and (E) and (F) represent phase contrast, fluorescein, and phycoerythrin epifluorescence illumination of the same microscopic fields, respectively. Note that the THϩ cells have a nonneuronal morphology. Not all progeny of a virus- infected founder cell express the transgene at this time point; thus, some THϩ cells appear mycϪ. Objective magnification, 20ϫ. In (G), the Phox2 Proteins Regulate Neurotransmitter Identity 697

Table 2. ERPH Blocks Transactivation of the TH Promoter by Phox2a in Transfected Fibroblasts Construct GFP ER ERPH

Experiment I 148.5 Ϯ 16.3 270 Ϯ 24 8.5 Ϯ 0.7 Experiment II ND 405 Ϯ 125 8.5 Ϯ 9 10T1/2 cells were transfected with TH-lacZ reporter plasmid, Phox2a expression plasmid, and the indicated constructs in a 1:2:4 ratio. The total number of ␤-galactosidase-positive cells per 35 mm dish is shown. The values represent the mean Ϯ range of two dishes for each experiment. Little or no expression of ␤-galactosidase was detected in cells transfected with the reporter plasmid alone (data not shown). In experiment I, staining of sister cultures with anti-myc tag antibody indicated a similar transfection efficiency in each case. Abbreviations: GFP, green fluorescent protein; ER, engrailed repressor domain alone; ERPH, engrailed repressor–Phox2a DNA binding domain fusion; and ND, not determined.

contrast, only 24.5% of GFP retrovirus–infected cells were THϩ (Figures 4B and 4D, green bar), a number not significantly different from the percentage of THϩ cells in the population as a whole (21.2% Ϯ 0.03%, n ϭ 28 fields scored; p Ͼ 0.05 by Student’s t test). These data suggested that the percentage of 2D10 cells that express TH can be raised by increasing the fraction of cells that express Phox2a above a certain threshhold. In contrast, essentially no 2D10 cells infected with the ERPH retrovirus expressed TH under the same conditions (Figures 4C, arrowheads, and 4D, blue bar). The fact that forced expression of wild-type Phox2a and ERPH yielded opposite effects on TH expression in 2D10 cells supports the idea that the dominant-negative mutant acts by interfering with endogenous Phox2a function in these cells.

Expression of Endogenous TH and DBH in Primary Figure 3. Phox2a-Infected Cells Express Both TH and c-RET Neural Crest Cells Is Induced The clone shown in (A) was double labeled with monoclonal antibody by BMP2 Plus Forskolin to c-RET (B) and TH (C) after growth in forskolin plus GDNF and We next wished to ask whether ERPH could also block Dex (Figure 2) for 96 hr. Arrows indicate cells coexpressing c-RET expression of endogenous TH in primary neural crest and TH. Note that the THϩ cells are a subset of the c-RETϩ cells. cells. This necessitated first identifying conditions for As all THϩ cells in such cultures expressed the retroviral Phox2a reliably inducing expression of TH in these cells, which transgene (see Figure 2), we infer that this clone was infected. Objec- had not been previously achieved. Prior experiments tive magnification, 20ϫ. had shown that BMP2 induces expression of endogenous Phox2a (but not TH) in NCSCs (Lo et al., 1998; A. The induction of TH in 2D10 cells by FGF plus Dex Groves et al., unpublished data). Since elevated intracel- was accompanied by a modest induction of Phox2a lular cAMP potentiated the induction of TH by retrovirally in a subset of the cells; Phox2b was not detectably expressed Phox2a (see Figure 2), we reasoned that addi- expressed (data not shown). Since only 20%–30% of tion of forskolin to BMP2-treated uninfected neural crest 2D10 cells express TH under these conditions, this result cultures might be sufficient to induce TH expression. suggested that levels of Phox2a might be limiting for Addition of forskolin to primary neural crest explants TH expression in these cells. In support of this, when grown in the standard concentration of BMP2 (10 ng/ 2D10 cells were infected with the wild-type Phox2a ret- ml) indeed induced TH expression, but only in a very rovirus, 60% of infected (mycϩ) cells expressed TH in small proportion of cells (data not shown). Since the FGF plus Dex (Figures 4A, arrows, and 4D, red bar). In biological effects of BMP family members are a sensitive

percentage of retrovirally infected clones (i.e., clones containing any Myc tagϩ cells) containing at least one THϩ cell was determined after 96 hr of growth under the indicated conditions. No expression of TH was observed under any conditions in GFP retrovirus–infected clones (green bars) but was detected in Phox2a-infected clones (red bars) and was potentiated by forskolin. (H) shows the time course of TH induction in NCSCs by Phox2a in the presence of 5 ␮M forskolin. In both (G) and (H), scale bar, mean Ϯ SEM. Asterisk indicates statistically significant differences between conditions or time points versus under control (“No add”) conditions, to either the 95% or 99% confidence levels as determined by Student’s t test. Neuron 698

of the crest cells were THϩ. The THϩ %12ف ,On average cells tended to be located at the center of the explants, where cell density is highest. Double labeling indicated that most or all of these THϩ cells coexpressed endogenous Phox2a (data not shown). Little or no expression of TH was obtained in cultures grown in forskolin in the absence of BMP2 (data not shown). Abundant expression of DBH was also detected under the same conditions (0.1 ng/ml BMP2 plus forskolin) (Figure 5C). Many of the DBHϩ cells coexpressed TH (Figures 5B and 5C, arrows); however, there were about as THϩ (%25ف) twice as many DBHϩ cells per explant cells. As was the case for TH, most or all DBHϩ cells coexpressed endogenous Phox2a (data not shown). The cells that coexpressed DBH and TH were process bearing (Figures 5B and 5C, arrows, and 6B), and such process-bearing THϩ or DBHϩ cells coexpressed neurofilament middle subunit (NF-M; Figures 6A, 6B, 6D, and 6F–6H). A second, rarer class of THϩ cell was also observed that had a flattened, fibroblastic morphology (Figures 5D and 5E). These flat THϩ cells did not coexpress DBH (Figure 5F) or NF-M (Figures 6C and 6E). In this respect, they resembled the flat THϩ cells observed following forced expression of Phox2a in NCSCs grown in the absence of BMP2 (Figure 2E). Unlike the case for TH, all DBHϩ cells appeared to coexpress NF-M and had a process-bearing morphology (Figures 6F–6H). These data indicated that BMP2 plus forskolin can induce expression of TH and DBH in neural crest cells. Little or no expression of these markers is detected in cultures containing forskolin alone or in cultures grown only in BMP2 (A. Groves and D. J. A., unpublished data). Expression of TH can be detected both in cells that coexpress neuronal markers and in nonneuronal cells, while that of DBH is restricted to cells expressing neuronal markers. The majority of THϩ, process-bearing cells appeared to coexpress DBH.

Endogenous Phox2 Proteins Are Necessary for Expression of TH and DBH as Well as for Neuronal Differentiation in Primary Neural Crest Cells We next asked whether Phox2a (or Phox2b) is required for the induction of TH and DBH in primary neural crest cells. When ERPH was expressed in neural crest cultures grown in the presence of 0.1 ng/ml BMP2 plus forksolin for 3 days, there was little if any overlap between cells expressing TH and those expressing the transgene (Figure 7C, arrowheads). Quantification indicated that Ͻ1% of all infected (mycϩ) cells were THϩ (Figure 7D, blue bar). A similarly low percentage of all Figure 4. Wild-Type and Dominant-Negative Phox2a Have Oppo- ϩ ϩ site Effects on TH Expression in 2D10 Cells TH cells were myc . By contrast, in control cultures infected with the GFP retrovirus, double labeled cells 2D10 cells were infected with the indicated retroviral constructs and analyzed for TH expression after 3 days of growth in FGF (10 ng/ could readily be seen (Figure 7A, arrows). Over 40% of ϩ ml) plus Dex (5 ␮M). TH is expressed in 25% of GFP-infected (mycϩ) all infected cells were TH in such cultures (Figure 7D, cells ([B]; [D], green bar) and in Ͼ60% of Phox2a-infected cells ([A]; red bar). Similar results were obtained when cells were [D], red bar); essentially no ERPH-infected cells express TH ([C]; infected with virus encoding the ER domain alone (Fig- [D], blue bar). ures 7B, arrows, and 7D, green bar). The frequency of GFP retrovirus–infected cells that were THϩ (41%) was similar to the frequency of THϩ cells in the cultures as function of their concentration, we tested lower and a whole (34%; p Ͼ 0.05 by Student’s t test). Conversely, higher concentrations of BMP2 in the presence of a the percentage of THϩ cells that were infected (33.5%) fixed amount of forskolin. At 0.1 ng/ml BMP2, a robust was similar to the percentage of total cells that were induction of TH expression was detected (Figure 5B). infected (27.2%; p Ͼ 0.05). These data indicate that Phox2 Proteins Regulate Neurotransmitter Identity 699

Figure 5. Induction of TH and DBH Expres- sion in Neural Crest Cells by BMP2 Plus For- skolin Primary neural crest explants were grown for 3 days in the presence of 0.1 ng/ml BMP2 plus 5 ␮M forskolin and double labeled with antibodies to TH (B and E) and DBH (C and F). At the center of the explant, where cell density is highest (A), many cells coexpress TH and DBH ([B and C], arrows), although the number of DBHϩ cells is higher. Note the short cytoplasmic extensions on the THϩ cells that coexpressed DBH ([B]; see also Fig- ures 6B and 6D). A second, rarer population of flat THϩ cell was detected (E) that did not express DBH (F). (A) and (D) are phase-contrast views of the fields shown in (B) and (C) and in (E) and (F), respectively. Objective magnification, 40ϫ.

retroviral infection itself does not change the frequency (data not shown). These data indicate that Phox2 proof TH expression and that THϩ cells are as equally in- teins are necessary for the neuronal differentiation in- fectable with control virus as THϪ cells. Thus, the failure duced by BMP2, even though they are not sufficient to to detect ERPH-infected cells expressing TH implies induce neuronal differentiation in the absence of BMP2 that expression of the dominant-negative Phox2a con- (Lo et al., 1998; Figures 2 and 3). struct blocks TH expression. Similar results were obtained in the case of DBH. In Discussion cultures infected with either of the control retroviruses, double labeled cells were easily detected (Figures 7E Using primary cultures of NCSCs, we have investigated and 7F, arrowheads, and 7H, red and green bars). By the specification of neurotransmitter identity by tran- contrast, when the ERPH virus was used, very little over- scription factors and environmental signals and its rela- lap was seen (Figure 7G, arrowheads); Յ2% of mycϩ tionship to neuronal differentiation. Evidence from both cells were DBHϩ (Figure 7H, blue bar), and a similarly gain- and loss-of-function experiments supports a di- low percentage of DBHϩ cells were mycϩ. As was the rect involvement of Phox2a (and/or Phox2b) in the regu- case for TH, the frequency of control GFP retrovirus– lation of TH and DBH gene expression in developing infected cells that were DBHϩ (43.5%) was similar to noradrenergic sympathetic neurons. In the presence of the frequency of DBHϩ cells in the cultures as a whole cAMP, Phox2a is sufficient to activate TH but not DBH. (36.7%; p Ͼ 0.05). Conversely, the percentage of DBHϩ Nevertheless, Phox2a/b is necessary for the induction cells that were infected (27.5%) was not significantly of both genes by BMP2 plus forskolin. Unexpectedly, different from the percentage of all cells that were in- Phox2 proteins are also required for BMP2-induced neu- fected (22.9%; p Ͼ 0.05). These data indicate that ex- ronal differentiation, although Phox2a is insufficient to pression of DBH, like that of TH, is also blocked by the promote such differentiation on its own. Thus, Phox2a/b dominant-negative Phox2a construct. participates in a regulatory circuit that integrates the While conducting these experiments, we noted that influence of multiple extracellular signals to coordinate crest cells infected with ERPH tended to have a flat- the expression of panneuronal and subtype-specific tened, nonneuronal morphology (Figures 8D and 8E). genes. Double labeling with neuronal markers such as NF160 confirmed that ERPH blocked the induction of neuronal Phox2 Proteins Are Direct Regulators differentiation by BMP2 in NCSC cultures, whether or of TH and DBH Expression not forskolin was added (Figures 8D–8F; data not Previous studies of TH and DBH transcription have pro- shown). In contrast, no such effect was observed with vided evidence that Phox2a and Phox2b can directly the GFP retrovirus (Figures 8A–8C) or with the ER virus regulate these genes. Both the TH and DBH promoters Neuron 700

Figure 6. TH Can Be Expressed with or without NF-M, while DBH Is Only Expressed To- gether with NF-M In (A) through (E), cultures were double labeled with anti-TH (red) and anti-NF-M (green). Double labeled cells are yellow. Sym- pathoadrenal progenitor–like THϩ cells that were process bearing coexpressed NF-M (A, B, and D), while the flat THϩ cells did not (C and E). In contrast, all of the DBHϩ cells (G) appeared to coexpress NF-M ([H], arrows; [F], arrowheads). (F) is a lower power double ex- posure of the field shown in (G) and (H). Ob- jective magnification, 20ϫ (A and F) and 40ϫ (B–E, G, and H).

contain binding sites for Phox2a/b (Zellmer et al., 1995), DBH expression in this system. Although targeted inacti- and this site is an important determinant of DBH pro- vation of Phox2a in mice did prevent transient expres- moter activity in PC12 cells (Shaskus et al., 1995). Fur- sion of DBH in cranial sensory ganglia, it did not prevent thermore, Phox2a/b can transactivate the TH and/or expression of TH by sympathetic neurons (Morin et al., DBH promoters (Zellmer et al., 1995; Kim et al., 1998; 1997). This difference is likely due to compensation by Yang et al., 1998). In the case of DBH, the DB1 (Phox2a/b Phox2b, which is coexpressed with Phox2a in sympa- binding) site has been shown to be critical for this trans- thetic neurons and other locations and which has an activation (Swanson et al., 1997; Kim et al., 1998; Yang identical DNA-binding domain (Pattyn et al., 1997). Re- et al., 1998). It is therefore likely that the homologous cent observations indicate that a Phox2b knockout pre- site is responsible for transactivation of the TH promoter vents the expression of both TH and DBH in developing by Phox2a as well (Zellmer et al., 1995). Taken together, sympathetic ganglia (A. Pattyn et al., unpublished data). these data strongly suggest that the regulation of endog- Thus, Phox2b can compensate for Phox2a but not vice enous TH and DBH gene expression by Phox2a/b in versa in the PNS. It will be interesting to determine NCSCs is direct. whether Phox2b has a gain-of-function phenotype iden- Our results extend these biochemical studies in sev- tical to that of Phox2a in NCSCs. eral important ways. First, we have assayed the activity Our data also demonstrate that forced expression of of Phox2a using the endogenous chromosomal TH and an exogenous transcription factor can induce neuro- DBH genes, rather than exogenous reporter constructs, transmitter enzyme gene expression in neural stem which contain only partial regulatory information from cells. This suggests that it may be possible to use tran- these genes. Second, we have analyzed Phox2a/b activ- scription factors to control the neurotransmitter pheno- ity in the cells in which it normally functions, rather than type of such cells for therapeutic purposes. However, in transformed nonneuronal cell lines. Third, we have it is likely that distinct transcriptional programs regulate tested not only the sufficiency but also the necessity of TH expression in central dopaminergic and peripheral Phox2a/b in the regulation of TH and DBH expression noradrenergic neurons (Gandelman et al., 1990; Ban- by employing a dominant-negative mutant. Finally, we erjee et al., 1992, 1994; Min et al., 1994; Wong et al., have examined the relationship of Phox2a/b function to 1995). the environmental signals that control neurotransmitter identity and neuronal differentiation in the neural crest. BMP2 and Intracellular cAMP Promote Expression Thus, our studies complement previous work by examin- of Noradrenergic Enzymes in Neural Crest Cells ing Phox2a/b function in its normal biological context. Previous work has shown that BMP family members The use of a dominant-negative form of Phox2a leaves are able to induce TH expression in avian neural crest uncertainty over which Phox2 protein(s) controls TH and cultures (Varley et al., 1995; Reissman et al., 1996; Varley Phox2 Proteins Regulate Neurotransmitter Identity 701

Figure 7. Dominant-Negative Phox2a Blocks TH and DBH Expression in Neural Crest Cells Grown in BMP2 Plus Forskolin Primary explants of rat neural crest cells were infected with GFP (A and E) or ERPH (C and G) viruses or with viruses containing the ER domain alone (B and F) and grown for 3 days in BMP2 plus forskolin. The cultures were double stained with antibodies to the Myc epitope tag (green) and TH ([A–C], red) or DBH ([E–G], red). Arrows in (A), (E), (B), and (F) indicate doubly labeled cells. Note the absence of overlap between ERPH-infected cells ([C and G], arrowheads) and THϩ (C) or DBHϩ (G) cells. Objective magnification, 40ϫ. (D and H) gives quantitation of the data in (A) through (C) and (E) through (G), respectively. The ordinate indicates the percentage of infected cells expressing TH (D) or DBH (H). The data represent the mean Ϯ SEM. Asterisk indicates significant difference by Student’s t test.

and Maxwell, 1996) and in embryos (Reissman et al., also modulate TH expression (Lewis et al., 1987; Yoon 1996). In clonal NCSC cultures, BMP2 alone was unable and Chikaraishi, 1992), but whether these control the to induce expression of TH (Shah et al., 1996), even initial induction of this gene is not clear. though it induces expression of endogenous Phox2a The idea that secondary environmental signals collab- (Lo et al., 1998) and Phox2b (L. L. and D. J. A., unpub- orate with BMP2/4 to specify noradrenergic neurotrans- lished data). cAMP potentiates transactivation of both mitter identity is consistent with extirpation experiments the TH and DBH promoters by Phox2a (Swanson et al., showing that both the notochord and dorsal aorta are 1997). Consistent with this, BMP2 plus forskolin dramati- required for catecholaminergic expression by chick cally induced both TH and DBH in primary crest cells. sympathetic precursors (Stern et al., 1991; Groves et al., The additional requirement of high cell density for TH 1995). The dorsal aorta expresses BMP2/4 (Reissman et and DBH expression further suggests that local cell–cell al., 1996; Shah et al., 1996). The notochord may therefore interactions may also be necessary for induction of be the source of a signal that collaborates with BMP2/4 these genes. to induce TH (Figure 9). Alternatively, the notochord BMP2 and forskolin likely exert their influences via could be required only to enhance BMP2/4 expression in interactions on the TH and DBH promoters between the dorsal aorta. Interestingly, the notochord produces Phox2a/b and proteins that bind a neighboring cAMP- norepinephrine (NE) (Strudel et al., 1977). NE promotes response element (CRE; Fung et al., 1992; Kim et al., TH and DBH expression in avian crest cultures, perhaps 1993; Swanson et al., 1997), such as CREB (Sheng and via elevated intracellular cAMP or calcium (Zhang and Greenberg, 1990; Montminy, 1997). This CRE is an im- Sieber-Blum, 1992; Zhang et al., 1997), which can acti- portant determinant of both TH (Kim et al., 1993; Lazaroff vate CREB (Sheng and Greenberg, 1990; Montminy, et al., 1995; Osaka and Sabban, 1997) and DBH (Ishiguro 1997). Multiple signals are also required for TH induction et al., 1995) transcription. Other promoter elements may in CNS dopaminergic neurons (Ye et al., 1998). Neuron 702

Figure 8. ERPH Inhibits Neuronal Differentia- tion of NCSCs Induced by BMP2 NCSCs infected with GFP (A–C) or ERPH (D–F) retroviruses were grown at clonal density in BMP2 for 4 days, fixed, and double stained with antibody to the Myc tag (B and E) and NF-M (C and F). (A) through (C) and (D) through (F) show phase contrast, brightfield, and epifluorescence views, respectively, of the same microscopic fields. Objective magnification, 40ϫ. Note the nonneuronal morphology of the ERPH-infected cells (D) and the lack of NF-M expression (F).

Relationship of Neurotransmitter expression that are induced by distinct but overlapping Identity to the Determination sets of environmental signals (Figure 9). of the Neuronal Phenotype This linkage appears critical in the case of DBH; unlike The present data demonstrate that Phox2a can activate TH, DBH expression cannot be activated by Phox2a in TH expression without inducing neuronal differentiation. the absence of neuronal differentiation. Consistent with We previously showed that, conversely, forced expres- this, in BMP2 plus forskolin, expression of DBH, unlike sion of MASH1 in NCSCs can induce neuronal differenti- that of TH, is never detected in nonneuronal cells (Figure ation without inducing expression of TH (Lo et al., 1998). 6). Thus, in contrast to TH, expression of DBH requires These data indicate that transcriptional control of TH both Phox2 proteins and the BMP2-dependent program expression can be experimentally uncoupled from that of neuronal differentiation (Figure 9). This latter require- of neuronal differentiation; furthermore, as shown here, ment may reflect the fact that the DBH gene contains it can even be uncoupled from that of DBH. These obser- a silencer element that represses its transcription in vations support the idea that generic neuronal differenti- nonneuronal cells (Ishiguro et al., 1993, 1995; Afar et al., ation and specification of neuronal identity may be con- 1996), similar to (but distinct from) the NRSE/RE-1 site trolled by separable genetic subprograms (Groves et (Schoenherr et al., 1996). Inactivation or repression of al., 1995; Anderson and Jan, 1997; Morin et al., 1997; the proteins that bind such a silencer, which is thought to Tanabe et al., 1998). accompany neuronal differentiation (Chong et al., 1995; Such subprograms must be coupled in some way in Schoenherr and Anderson, 1995; Chen et al., 1998), would order to coordinate the expression of subtype-specific then be required to permit expression of DBH. Alterna- and panneuronal properties. Unexpectedly, we found tively, DBH expression could require positive neuro- that Phox2 proteins are necessary although not suffi- genic regulators in addition to Phox2a/b (Figure 9). cient (Lo et al., 1998) to activate the expression of Why should expression of TH and DBH be linked in panneuronal genes. It is unlikely that ERPH artifactually different ways to the neurogenic subprogram if they inhibits neurogenesis, since expression of panneuronal function in the same neurotransmitter biosynthetic path- markers such as neurofilament is also lost in the auto- way? Perhaps it reflects the fact that DBH is a membrane nomic ganglia of Phox2b knockouts (A. Pattyn et al., protein localized in synaptic vesicles, special organelles unpublished data). This requirement could be direct or whose assembly requires many neuron-specific gene indirect; the presence of Phox2-binding sites in the pro- products. By contrast, TH is a soluble, cytosolic enzyme. moter of NCAM, a panneuronal gene (Valarche´ et al., Expression of DBH may thus be permitted only in coordi- 1993), supports the latter possibility. Thus, Phox2 pro- nation with that of generic neuronal proteins that are teins appear to be independently required for both required for its proper subcellular localization. panneuronal and neurotransmitter enzyme gene expression in sympathetic neuroblasts. However, panneuronal genes can be induced by BMP2 alone (Shah et al., 1996; A Regulatory Pathway for the Specification Figure 8), while induction of TH and DBH requires both of Neurotransmitter Identity BMP2 and cAMP. In this way, Phox2a/b links subpro- Although many other factors are almost certainly in- grams for both panneuronal and subtype-specific gene volved, it is now possible to sketch an outline of the Phox2 Proteins Regulate Neurotransmitter Identity 703

replating. Clones were considered infected if they contained any Myc tagϩ cells, although not all progeny of infected cells express the transgene (Lo et al., 1998). Primary neural crest explants were obtained as previously described (Lo et al., 1998). Three rounds of retroviral infection were performed on six neural crest explants that had been plated in close vicinity and grown for 16 hr after scraping the neural tubes away. BMP2 and forskolin (5 ␮M) were added 16 hr after infection, and cells were fixed 3 days later. To determine the percentage of TH, DBH, or Myc tagϩ cells, 20ϫ pictures were taken near the center of the explants, in the region containing the most TH- or DBH-positive cells. Counts were obtained directly from photographs of randomly selected fields. 2D10 cells were maintained as described (Lo et al., 1990). For cells were plated overnight in the center 5000ف ,retroviral infection of a fibronectin-coated 35 mm dish, subjected to three rounds of viral infection, and replated at 800–1000 cells per 35 mm dish on fibronectin. FGF and Dex were added 16–18 hr after replating, and cultures were fixed 3 days after replating. Forskolin and Dex were purchased from Sigma and used at 5 ␮M. Basic FGF was obtained from R and D and used at 10 ng/ml. GDNF was purchased from Promega and used at 50 ng/ml.

Plasmid Constructions Plasmid pERPH contains the homeodomain of Phox2a (Valarche´ et al., 1993; Pattyn et al., 1997) fused to the Drosophila ER domain (Han and Manley, 1993) in pBluescript KS-II. A PCR-amplified coding Figure 9. Schematic Illustrating Regulation and Function of Phox2 fragment was subcloned into a modified pBABE vector (Lo et al., Proteins in Specifying Different Components of the Noradrenergic 1997) to generate pBABE–ERPH. To generate pBABE–GFP, the se- Sympathetic Phenotype quence encoding a humanized GFP was amplifed from pGreen Lan- Arrows indicate direct or indirect relationships, thick arrows likely tern (Gibco/Life Technologies) and subcloned into a modified direct relationships. Phox2b is expressed independent of MASH-1 pBABE vector that contains a nuclear localization signal fused in- (Hirsch et al., 1998) but is also induced by BMPs. The notochord is frame to a Myc epitope tag (L. Sommer and D. J. A., unpublished required for induction of TH (Groves et al., 1995) and could provide a data). To generate pBABE–ER, the Drosophila ER domain was ampli- cAMP-inducing signal or promote BMP2/4 expression by the dorsal fied from pERPH with primers containing an in-frame stop codon aorta (Reissman et al., 1996). DBH (but not TH) activation by at the 3Ј end and subcloned into the modified pBABE vector. All Phox2a/b requires a neurogenic program, either to extinguish hypo- PCR amplifications used pfu polymerase. The pTH–lacZ reporter thetical silencer factors (N-Sil) (Ishiguro et al., 1995) or to activate construct was generated by cloning a 1.5 kb EcoRI–NaeI fragment enhancer factors. c-RET can be activated by Phox2a in the absence from the rat TH gene (Banerjee et al., 1992) 5Ј to a 4.7 kb HindIII–ScaI of cAMP or neuronal differentiation (Lo et al., 1998). The convergent fragment from pRSV-lacZ (Johnson et al., 1992), containing the lacZ actions of Phox2a and Phox2b represent similar activities, not heter- gene and an SV40 3Ј untranslated region cassette, into pGEM3. All odimer formation. Phox2b also activates Phox2a (A. Pattyn et al., constructs were confirmed by sequencing. unpublished data). Transactivation Assay regulatory pathway that controls the specification of the 10T1/2 cells were plated into 35 mm dishes at medium density in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine neuronal fate and neurotransmitter identity in sympa- serum for 6 hr and then transfected by a modified calcium phosphate thetic neurons (Figure 9). BMP2 (or BMP4), expressed procedure with 0.2 ml of CaPO4/DNA mixture per plate. Each dish in the dorsal aorta (Reissman et al., 1996; Shah et al., received 0.8 ␮gofpTH-lacZ, 1.6 ␮g of pBABE–Phox2a, and 3.2 ␮g 1996), acts on neighboring neural crest–derived cells to of either pBABE–GFP, pBABE–ER, or pBABE–ERPH. After 14 hr, induce expression of MASH1 (Shah et al., 1996) and in the dishes were washed, refed, and incubated for 36 hr prior to parallel Phox2b (L. L. and D. J. A., unpublished data), fixation in formaldehyde-glutaraldehyde. Cells were developed in whose expression is independent of MASH1 (Hirsch et X-gal for 6 hr at room temperature. Detailed fixation and development protocols are available on request. al., 1998). MASH1 in turn (directly or indirectly) induces expression of Phox2a (Lo et al., 1998). Phox2a and/or Immunocytochemistry Phox2b bind to the TH and DBH promoters to activate Monoclonal antibodies to c-RET (Lo and Anderson, 1995), MASH1 their transcription in collaboration with CREB (or other (Lo et al., 1991), human c-Myc (Evan et al., 1985), and polyclonal CRE-binding proteins), which is activated by signals that anti-Phox2a (Tiveron et al., 1996) and -Phox2b (Pattyn et al., 1997) elevate intracellular cAMP or calcium (Figure 9). In paral- have been described previously. Monoclonal anti-DBH (Chemicon lel, MASH1 activates a program of generic neuronal dif- and Pharmingen), anti-TH (Boehringer-Mannheim), and rabbit anti- ferentiation (Sommer et al., 1995; Lo et al., 1998), for NF-M (Chemicon) were used at 1:500. Double labeling with anti-TH which Phox2 proteins are also necessary. This neuro- (mouse IgG2a) versus anti–Myc tag (mouse IgG1) or DBH (mouse IgG1) was performed using isotype-specific fluorescent secondary genic program is required in addition to Phox2a/b for antibodies (Southern Biotechnology). Double labeling with mouse expression of DBH. In this way, Phox2 proteins form a anti-DBH versus rabbit anti–Myc tag (Santa Cruz Biotechnology) node in a regulatory network that links environmental was performed using species-specific secondary antibodies (Jack- signals that specify neurotransmitter identity to those son ImmunoResearch). that determine the neuronal fate. Acknowledgments Experimental Procedures We acknowledge Andrew Groves for initial studies of TH induction; Cell Cultures and Retroviral Infection Christo Goridis for support and input; members of the Anderson lab Neural crest cultures and retroviral infection were performed essen- for helpful discussions; L. Wang, S. Padilla, and L. Dinh for technical tially as described (Lo et al., 1998). Factors were added 1–2 hr after support; G. Mosconi for lab management; L. Sommer for the GFP Neuron 704

construct; and Alice Paquette for a critical reading of the manuscript. Han, K., and Manley, J.L. (1993). Functional domains of the Drosoph- This work was supported in part by the Association pour la Recher- ila engrailed protein. EMBO J. 12, 2723–2733. che sur le Cancer to J.-F. B. D. J. A. is an Investigator of the Howard Hirsch, M.R., Tiveron, M.C., Guillemot, F., Brunet, J.F., and Goridis, Hughes Medical Institute. C. (1998). Control of noradrenergic differentiation and Phox2a expression by MASH1 in the central and peripheral nervous system. Development 125, 599–608. Received September 3, 1998; revised February 23, 1999. Ishiguro, H., Kim, K.T., and Kim, K.-S. (1993). Neuron-specific expression of the human dopamine-␤-hydroxylase gene requires both References the cAMP-response element and a silencer region. J. Biol. Chem. 268, 17987–17994. Afar, R., Silverman, R., Aguanno, A., and Albert, V.R. (1996). Positive Ishiguro, H., Kim, K.S., and Joh, T.H. (1995). Identification of a nega- and negative elements contribute to the cell-specific expression of tive regulatory element in the 5Ј-flanking region of the human dopa- the rat dopamine ␤-hydroxylase gene. Mol. Brain Res. 36, 79–92. mine ␤-hydroxylase gene. Mol. Brain Res. 34, 251–261. Anderson, D.J., and Jan, Y.N. (1997). The determination of the neu- Johnson, J.E., Birren, S.J., and Anderson, D.J. (1990). Two rat homo- ronal phenotype. In Molecular and Cellular Approaches to Neural logues of Drosophila achaete-scute specifically expressed in neu- Development, W.M. Cowan, ed. (New York: Oxford University Press), ronal precursors. Nature 346, 858–861. pp 26–63. Johnson, J.E., Birren, S.J., Saito, T., and Anderson, D.J. (1992). Badiani, P., Corbella, P., Kioussis, D., Marvel, J., and Weston, K. DNA binding and transcriptional regulatory activity of mammalian (1994). Dominant interfering alleles define a role for c-Myb in T-cell achaete-scute homologous (MASH) proteins revealed by interaction development. Genes Dev. 8, 770–782. with a muscle-specific enhancer. Proc. Natl. Acad. Sci. USA 89, Banerjee, S.A., Hoppe, P., Brilliant, M., and Chikaraishi, D.M. (1992). 3596–3600. 5Ј flanking sequences of the rat tyrosine hydroxylase gene target Kim, H.S., Seo, H., Yang, C., Brunet, J.-F., and Kim, K.-S. (1998). accurate tissue-specific, developmental, and transsynaptic expres- Noradrenergic-specific transcription of the dopamine ␤-hydroxylase sion in transgenic mice. J. Neurosci. 12, 4460–4467. gene requires synergy of multiple cis-acting elements including at Banerjee, S.A., Roffler-Tarlov, S., Szabo, M., Frohman, L., and Chi- least two Phox2a-binding sites. J. Neurosci. 18, 8247–8260. karaishi, D.M. (1994). DNA regulatory sequnces of the rat tyrosine Kim, K.S., Lee, M.K., Carroll, J., and Joh, T.H. (1993). Both the basal hydroxylase gene direct correct catecholaminergic cell-type speci- and inducible transcription of the tyrosine hydroxylase gene are ficity of a human growth hormone reporter in the CNS of transgenic dependent upon a cAMP response element. J. Biol. Chem. 268, mice causing a dwarf phenotype. Brain Res. Mol. Brain Res. 24, 15689–15695. 89–106. Lazaroff, M., Patankar, S., Yoon, S.O., and Chikaraishi, D.M. (1995). Bellefroid, E.J., Bourguignon, C., Hollemann, T., Ma, Q., Anderson, The cyclic AMP response element directs tyrosine hydroxylase ex- D.J., Kintner, C., and Pieler, T. (1996). X-MyT1, a Xenopus C2HC- pression in catecholaminergic central and peripheral nervous sys- type zinc finger protein with a regulatory function in neuronal differ- tem cell lines from transgenic mice. J. Biol. Chem. 270, 21579–21589. entiation. Cell 87, 1191–1202. Lewis, E.J., Harrington, C.A., and Chikaraishi, D.M. (1987). Transcrip- Birren, S.J., and Anderson, D.J. (1990). A v-myc-immortalized sym- tional regulation of the tyrosine hydroxylase gene by glucocorticoid pathoadrenal progenitor cell line in which neuronal differentiation and cyclic AMP. Proc. Natl. Acad. Sci. USA 84, 3550–3554. is initiated by FGF but not NGF. Neuron 4, 189–201. Lindsay, R.M., and Yancopoulos, G.D. (1996). GDNF in a bind with Chen, Z.-F., Paquette, A.J., and Anderson, D.J. (1998). NRSF/REST known orphan: accessory implicated in new twist. Neuron 17, is required in vivo for repression of multiple neuronal target genes 571–574. during embryogenesis. Nat. Genet. 20, in press. Lo, L., Sommer, L., and Anderson, D.J. (1997). MASH1 maintains Chong, J.A., Tapia-Ramirez, J., Kim, S., Toledo-Aral, J.J., Zheng, competence for BMP2-induced neuronal differentiation in post- Y., Boutros, M.C., Altshuller, Y.M., Frohman, M.A., Kraner, S.D., and migratory neural crest cells. Curr. Biol. 7, 440–450. Mandel, G. (1995). REST: a mammalian silencer protein that restricts Lo, L., Tiveron, M.-C., and Anderson, D.J. (1998). MASH1 activates sodium channel expression to neurons. Cell 80, 949–957. expression of the paired homeodomain transcription factor Phox2a, and couples pan-neuronal and subtype-specific components of au- Edlund, T., and Jessell, T.M. (1999). Progression from extrinsic to tonomic neuronal identity. Development 125, 609–620. intrinsic signaling in cell fate specification: a view from the nervous system. Cell 96, 211–224. Lo, L.-C., and Anderson, D.J. (1995). Postmigratory neural crest cells expressing c-ret display restricted developmental and proliferative Evan, G.I., Lewis, G.K., Ramsay, G., and Bishop, J.M. (1985). Isola- capacities. Neuron 15, 527–539. tion of monoclonal antibodies specific for human c-myc proto-onco- gene product. Mol. Cell. Biol. 5, 3610–3616. Lo, L.-C., Birren, S.J., and Anderson, D.J. (1990). v-myc immortaliza- tion of early rat neural crest cells yields a clonal cell line which Fung, B.P., Yoon, S.O., and Chikaraishi, D.M. (1992). Sequences generates both glial and adrenergic progenitor cells. Dev. Biol. 145, that direct rat tyrosine hydroxylase gene expression. J. Neurochem. 139–153. 58, 2044–2052. Lo, L.-C., Johnson, J.E., Wuenschell, C.W., Saito, T., and Anderson, Furukawa, T., Morrow, E.M., and Cepko, C.L. (1997). Crx, a novel D.J. (1991). Mammalian achaete-scute homolog 1 is transiently ex- otx-like homeobox gene, shows photoreceptor-specific expression pressed by spatially-restricted subsets of early neuroepithelial and and regulates photoreceptor differentiation. Cell 91, 531–541. neural crest cells. Genes Dev. 5, 1524–1537. Gandelman, K.Y., Coker, G.T., Moffat, M., and O’Malley, K.L. (1990). Mandel, G., and McKinnon, D. (1993). Molecular basis of neural- Species and regional differences in the expression of cell-type spe- specific gene expression. Annu. Rev. Neurosci. 16, 323–345. cific elements at the human and rat tyrosine hydroxylase gene loci. Maxwell, G.D., Reid, K., Elefanty, A., Bartlett, P.F., and Murphy, J. Neurochem. 55, 2149–2152. M. (1996). Glial cell line-derived neurotrophic factor promotes the Goridis, C., and Brunet, J.-F. (1999). Transcriptional control of neuro- development of adrenergic neurons in mouse neural crest cultures. transmitter phenotype. Curr. Opin. Neurobiol. 9, 47–53. Proc. Natl. Acad. Sci. USA 93, 13274–13279. Groves, A.K., George, K.M., Tissier-Seta, J.-P., Engel, J.D., Brunet, Min, N., Joh, T.H., Kim, K.S., Peng, C., and Song, J.H. (1994). 5Ј J.-F., and Anderson, D.J. (1995). Differential regulation of transcrip- upstream DNA sequence of the rat tyrosine hydroxylase gene directs tion factor gene expression and phenotypic markers in developing high-level and tissue-specific expression to catecholaminergic neu- sympathetic neurons. Development 121, 887–901. rons in the central nervous system of transgenic mice. Brain Res. Guillemot, F., Lo, L.-C., Johnson, J.E., Auerbach, A., Anderson, D.J., Mol. Brain Res. 27, 281–289. and Joyner, A.L. (1993). Mammalian achaete-scute homolog-1 is Mitani, S., Du, H., Hall, D.H., Driscoll, M., and Chalfie, M. (1993). required for the early development of olfactory and autonomic neu- Combinatorial control of touch receptor neuron expression in rons. Cell 75, 463–476. Caenorhabditis elegans. Development 119, 773–783. Phox2 Proteins Regulate Neurotransmitter Identity 705

Montminy, M. (1997). Transcriptional regulation by cyclic AMP. C., and Brunet, J.-F. (1993). The mouse homeodomain protein Phox2 Annu. Rev. Biochem. 66, 807–822. regulates NCAM promoter activity in concert with Cux/CDP and is Morin, X., Cremer, H., Hirsch, M.-R., Kapur, R.P., Goridis, C., and a putative determinant of neurotransmitter phenotype. Development Brunet, J.-F. (1997). Defects in sensory and autonomic ganglia and 119, 881–896. absence of locus coeruleus in mice deficient for the homeobox gene Varley, J.E., and Maxwell, G.D. (1996). BMP-2 and BMP-4, but not Phox2a. Neuron 18, 411–423. BMP-6, increase the number of adrenergic cells which develop in Osaka, H., and Sabban, E.L. (1997). Requirement for cAMP/calcium quail trunk neural crest cultures. Exp. Neurol. 140, 84–94. response element but not AP-1 site in fibroblast growth factor-2- Varley, J.E., Wehby, R.G., Rueger, D.C., and Maxwell, G.D. (1995). elicited activation of tyrosine hydroxylase gene expression in PC12 Number of adrenergic and islet-1 immunoreactive cells is increased cells. Brain Res. Mol. Brain Res. 49, 222–228. in avian trunk neural crest cultures in the presence of human recom- Pattyn, A., Morin, X., Cremer, H., Goridis, C., and Brunet, J.-F. (1997). binant osteogenic protein-1. Dev. Dynam. 203, 434–447. Expression and interactions of the two closely related homeobox Wong, S.C., Moffat, M.A., Coker, G.T., Merlie, J.P., and O’Malley, genes Phox2a and Phox2b during neurogenesis. Development 124, K.L. (1995). The 3Ј flanking region of the human tyrosine hydroxylase 4065–4075. gene directs reporter gene expression in peripheral neuroendocrine Quinn, J.P. (1996). Neuronal-specific gene expression—the interac- tissues. J. Neurochem. 65, 23–31. tion of both positive and negative transcriptional regulators. Prog. Yang, C.Y., Kim, H.S., See, H., Kim, C.H., Brunet, J.F., and Kim, K.S. Neurobiol. 50, 363–379. (1998). Paired-like homeodomain proteins, Phox2a and Phox2b, are Reissman, E., Ernsberger, U., Francis-West, P.H., Rueger, D., Brick- responsible for noradrenergic cell–specific transcription of the do- ell, P.D., and Rohrer, H. (1996). Involvement of bone morphogenetic pamine ␤-hydroxylase gene. J. Neurochem. 71, 1813–1826. protein-4 and bone morphogenetic protein-7 in the differentiation Ye, W., Shimamura, K., Rubenstein, J.L., Hynes, M.A., and Rosen- of the adrenergic phenotype in developing sympathetic neurons. thal, A. (1998). FGF and Shh signals control dopaminergic and sero- Development 122, 2079–2088. tonergic cell fate in the anterior neural plate. Cell 93, 755–766. Schoenherr, C.J., and Anderson, D.J. (1995). The neuron-restrictive Yoon, S.O., and Chikaraishi, D.M. (1992). Tissue-specific transcrip- silencer factor (NRSF): a coordinate repressor of multiple neuron- tion of the rat tyrosine hydroxylase gene requires synergy between specific genes. Science 267, 1360–1363. an AP-1 motif and an overlapping E box–containing dyad. Neuron Schoenherr, C.J., Paquette, A.J., and Anderson, D.J. (1996). Identifi- 9, 55–67. cation of potential target genes for the neuron-restrictive silencer Zellmer, E., Zhang, Z., Greco, D., Rhodes, J., Cassel, S., and Lewis, factor. Proc. Natl. Acad. Sci. USA 93, 9881–9886. E.J. (1995). A homeodomain protein selectively expressed in norad- Shah, N.M., Groves, A., and Anderson, D.J. (1996). Alternative neural renergic tissue regulates transcription of neurotransmitter biosyn- crest cell fates are instructively promoted by TGF␤ superfamily thetic genes. J. Neurosci. 15, 8109–8120. members. Cell 85, 331–343. Zhang, J.-M., and Sieber-Blum, M. (1992). Characterization of the Sharma, K., Sheng, H.Z., Lettieri, K., Li, H., Karavanat, A., Potter, norepinephrine uptake system and the role of norepinephrine in the S., Westphal, H., and Pfaff, S. (1998). LIM homeodomain factors expression of the adrenergic phenotype by quail neural crest cells Lhx3 and Lhx4 assign subtype identities for motor neurons. Cell 95, in clonal culture. Brain Res. 570, 251–258. 817–828. Zhang, J.-M., Dix, J., Langtimm-Sedlak, C.J., Trusk, T., Schroeder, Shaskus, J., Zellmer, E., and Lewis, E.J. (1995). A negative regulatory B., Hoffmann, R., Strosberg, A.D., Winslow, J.W., and Sieber-Blum, element in the rat dopamine ␤-hydroxylase gene contributes to the M. (1997). Neurotrophin-3 and norepinephrine-mediated adrenergic cell type specificity of expression. J. Neurochem. 64, 52–60. differentiation and the inhibitory action of desipramine and cocaine. Sheng, M., and Greenberg, M.E. (1990). The regulation and function J. Neurobiol. 32, 262–280. of c-fos and other immediate early genes in the nervous system. Neuron 4, 477–485. Note Added in Proof Smith, J., and Fauquet, M. (1984). Glucocorticoids stimulate adrenergic differentiation in cultures of migrating and premigratory neural The data referred to throughout as “A. Pattyn et al., unpublished crest. J. Neurosci. 4, 2160–2172. data” are now in press: Pattyn, A., Morin, X., Cremer, H., Goridis, C., and Brunet, J.-F. (1999). The homeobox gene Phox2b is essential Sommer, L., Shah, N., Rao, M., and Anderson, D.J. (1995). The cellular function of MASH1 in autonomic neurogenesis. Neuron 15, 1245– for the development of all autonomous derivatives of the neural 1258. crest. Nature, in press. Stemple, D.L., and Anderson, D.J. (1992). Isolation of a stem cell for neurons and glia from the mammalian neural crest. Cell 71, 973–985. Stern, C.D., Artinger, K.B., and Bronner-Fraser, M. (1991). Tissue interactions affecting the migration and differentiation of neural crest cells in the chick embryo. Development 113, 207–216. Strudel, G., Recasens, M., and Mandel, P. (1977). Identification de catecholamines et de serotonine dans les chordes d’embyrons de poulet. Compt. Rend. Acad. Sci. (Paris) 284, 967–969. Swanson, D.J., Zellmer, E., and Lewis, E.J. (1997). The homeodomain protein Arix interacts synergistically with cyclic AMP to regulate expression of neurotransmitter biosynthetic genes. J. Biol. Chem. 272, 27382–27392. Tanabe, Y., William, C., and Jessell, T.M. (1998). Specification of motor neuron identity by the MNR2 homeodomain protein. Cell 95, 67–80. Thor, S., and Thomas, J.B. (1997). The Drosophila islet gene governs axon pathfinding and neurotransmitter identity. Neuron 18, 397–409. Tiveron, M.-C., Hirsch, M.-R., and Brunet, J.-F. (1996). The expression pattern of the transcription factor Phox2a delineates synaptic pathways of the autonomic nervous system. J. Neurosci. 16, 7649– 7660. Valarche´ , I., Tissier-Seta, J.-P., Hirsch, M.-R., Martinez, S., Goridis,