Gene regulatory logic of dopamine neuron differentiation.

Dopamine signalling regulates a variety of complex behaviours, and defects in dopamine neuron function or survival result in severe human pathologies, such as Parkinson's disease. The common denominator of all dopamine neurons is the expression of dopamine pathway genes, which code for a set of phylogenetically conserved proteins involved in dopamine synthesis and transport. Gene regulatory mechanisms that result in the direct activation of dopamine pathway genes and thereby ultimately determine the identity of dopamine neurons are poorly understood in all systems studied so far. Here we show that a simple cis-regulatory element, the dopamine (DA) motif, controls the expression of all dopamine pathway genes in all dopaminergic cell types in Caenorhabditis elegans. The DA motif is activated by the ETS transcription factor AST-1. Loss of ast-1 results in the failure of all distinct dopaminergic neuronal subtypes to terminally differentiate. Ectopic expression of ast-1 is sufficient to activate the dopamine pathway in some cellular contexts. Vertebrate dopamine pathway genes also contain phylogenetically conserved DA motifs that can be activated by the mouse ETS transcription factor Etv1 (also known as ER81), and a specific class of dopamine neurons fails to differentiate in mice lacking Etv1. Moreover, ectopic Etv1 expression induces dopaminergic fate marker expression in neuronal primary cultures. Mouse Etv1 can also functionally substitute for ast-1 in C. elegans. Our studies reveal a simple and apparently conserved regulatory logic of dopamine neuron terminal differentiation and may provide new entry points into the diagnosis or therapy of conditions in which dopamine neurons are defective.

Nervous systems generally harbor distinct populations of dopaminergic (DA) neurons that derive from different precursor cells. Despite their diverse origin, all DA neurons share the expression of a core set of 5 genes that code for enzymes and transporters which synthesize, package and re-uptake dopamine (“dopamine pathway genes”; Fig.1a). The regulatory logic of the terminal differentiation of DA neurons, manifested by the induction of the DA pathway genes, can, in theory, be described by two distinct models. In model #1, each dopamine pathway gene is independently activated by a distinct set of regulatory factors and, as a reflection of their distinct developmental history, each DA neuron subtype utilizes a distinct set of regulatory molecules (Fig. 1b). In model #2, each dopamine pathway gene is regulated by the same regulatory factor(s) and those factor(s) are the same in each DA neuron subtype (Fig. 1b). These two models make specific predictions about the cis-regulatory architecture of dopamine pathway genes. In model #1, each dopamine pathway gene is controlled by distinct cis-regulatory motifs and different motifs are active in individual DA neuron subtypes. In model #2, there is a single motif for all pathway gene that is utilized in all different DA neuron subtype (Fig. 1b). To test these models, we made use of the dopaminergic system of the nematode C. elegans, which contains 4 distinct, lineally unrelated classes of dopaminergic neurons that express the same set of highly conserved dopamine pathway genes (Fig. 1c) 3. We systematically dissected the cis-regulatory regions of all DA pathways genes in the context of gfp reporters expressed in transgenic worms (Fig. 1c-h; Suppl. Fig. S1). The cis-regulatory analysis of two genes exclusively expressed in the DA neurons, the dopamine transporter gene dat-1/DAT and the tyrosine hydroxylase gene cat-2/TH, reveals the existence of a small cis-regulatory module (CRM) in each promoter that is required and sufficient to drive expression in all DA neurons (Fig. 1d and e). Dopamine pathway genes expressed in both DA and serotonergic (5-HT) neurons (cat-1/VMAT, cat-4/GTPCH, bas-1/AAA) contain separable CRMs for expression in DA and 5-HT neurons (Fig. 1f-h).

Fig. 1

Characterization of the DA motif in C. elegans

(a) Schematic representation of a DA neuron synapse. AAAD: aromatic L-amino acid decarboxylase; bas-1: Biogenic Amine Synthesis related 1; cat: abnormal CATecholamine distribution; DA: dopamine; dat: DopAmine Transporter; GTPCH: GTP cyclohydrolase; TH: Tyrosine hydroxylase; Tyr: tyrosine; VMAT: vesicular monoamine transporter.

(b) Schematic representation of two different models for DA terminal differentiation. See text for explanations.

(c) Picture of an adult worm expressing GFP under the control of the full length dat-1/DAT promoter, labeling all C. elegans DA neurons. Similarly, cat-2/TH is also exclusively expressed in DA neurons (not shown), as C.elegans contains no adrenergic or noradrenergic neurons.

(d) dat-1/DAT promoter analysis. Schematic representation of the dat-1/DAT locus with its upstream region. Exons are represented as red blocks, the upstream gene is shown in grey. Below: representation of cloned and injected constructs, and expression pattern in the DA and serotonergic (5-HT) neurons. Thick black lines symbolize the promoter piece placed in front of GFP (green box). Red cross represents a mutated EBS. “+” indicates >10% penetrant expression in more than half of the transgenic lines examined; “+/−“ also means >10% penetrant expression, but the penetrance is lower than in the corresponding full-length construct, “-” indicates <10% penetrant expression in more than half of the transgenic lines examined, “n.d” means not determined.

(e-h) Analysis of the regulatory regions of all other dopamine pathway genes. “+*” means dimmer gfp expression than corresponding wild type construct. See Fig. S1 and S3 for all primary data and nature of the mutations.

(i) The sequence alignment of all functional EBSs defines a position weight matrix (PWM) that is represented by a sequence logo. The conserved core in all sequences constitutes the DA motif. See Fig. S5 for sequences used to define the DA motif.

The DA-specific CRM of the dat-1/DAT locus contains a small sequence motif that is conserved in three other Caenorhabditis species (Fig. 1d; Suppl. Fig. S5) and through mutation was found to be required for dat-1 expression in all DA neurons (Fig. 1d). This motif is also sufficient to drive expression in all DA neurons, either when tested in isolation or when appended to the CRM of another neuron-specific gene (Suppl. Fig. S2a). Bioinformatics analysis predicts the binding of six different types of transcription factors to this conserved motif (Suppl. Fig. S2b). Point mutations that specifically abolish the predicted binding of some factors while keeping others intact reveal that the only predicted motif that can be made responsible for cis-regulatory motif activity in the DA neurons is a predicted ETS transcription factor binding site (EBS) defined by an invariant GGAW core sequence (Suppl. Fig. S2b). The DA-expressed CRM of all other dopamine pathway genes also contain predicted EBSs and mutational analysis corroborates their requirement for the correct expression in all DA neurons of C. elegans hermaphrodites (Fig.1f-h; Suppl. Fig.S3) and in the three additional DA pairs present in the male (Suppl. Fig. S4). All the functionally characterized EBSs are conserved in other Caenorhabditis species, they can occur in either orientation and at different distances from the transcriptional start (Fig. S5). The weight matrix generated with all these sequences defines a consensus EBS sequence motif that we term the “DA motif” (Fig.1i; Suppl. Fig.S5).

Analyzing the expression of the DA marker dat-1::gfp in mutants that lack each of the ten C. elegans ETS family members (Fig. S6), we find that loss that all ets mutants showed wild-type dat-1::gfp expression except for animals lacking the Axon STeering defect-1 (ast-1) gene, previously identified as a gene controlling axon outgrowth in the ventral nerve cord 4. Moreover, we found that a mutant allele, ot417, that we retrieved from an unbiased forward genetic screen for mutants in which DA fate is inappropriately executed 5, is an allele of ast-1 (Fig.2a). The expression of all five dopamine pathway genes is strongly affected if not completely lost in ast-1 mutants (Fig.2b,c; Suppl. Table S1; Suppl. Fig.S7). Two other DA terminal differentiation markers, the ion channels asic-1 6 and trp-4 7, also fail to be expressed in the DA neurons of ast-1 mutants (Suppl. Fig. S8). Both genes contain phylogenetically conserved DA motifs in their regulatory regions. ast-1 therefore appears to affect DA fate broadly, which is further corroborated by axon pathfinding defects we observe in ast-1 mutants (Suppl. Fig. S9). Loss of DA marker gene expression is not a reflection of early lineage specification defects and/or absence of the neurons, as assessed by analysis of additional fate marker (Suppl. Fig. S8).

Fig. 2

ast-1 is required to induce and maintain DA neuron differentiation

(a) Schematic representation of ast-1 locus and the mutants available for this gene.

(b) Representative example of loss of DA fate marker in ast-1 mutants. See Fig. S7 and S8 for other examples and Table S1 for quantification of data.

(c) Summary of ast-1 null mutant phenotype. +: fate marker expressed; -: fate marker not expressed. Due to early larval lethality, only the embryonically generated DA head neurons, but not the postembryonically generated PDE neurons could be scored for developmental defects in ast-1 null mutants. Markers that were expressed in both DA and 5-HT neurons were assayed with an rfp reporter in a transgenic background in which 5-HT neurons were labeled with gfp (Is[tph-1::gfp) so as to allow for loss of expression specifically in the DA neurons. See Fig. S7 for primary data.

(d) Expression of an ast-1::yfp reporter gene 4 in DA neurons. DA neurons are labeled with dat-1::mCherry. Scale bar: 10 μm.

(e) Rescue of the ast-1 mutant phenotype. We used a hypomorphic allele, hd-1, in which dat-1 expression is unaffected (Table S1), to drive ast-1 or a mouse homolog, etv-1, under control of the DA-specific dat-1 promoter and assayed expression of cat-2::gfp (otIs199).

(f) Developmentally staged ast-1(rh300) animals, containing the heat shock-inducible ast-1 array otIs198 and the DA fate marker cat-2::gfp (otIs199), were grown under non-inducible condition to the first larval stage (resulting in an absence of cat-2::gfp expression in 100% of animals); ast-1 was then induced by heat shock at the L1 stage and animals scored 4 hours and 3 days after heat shock. Of the 40 animals found to turn on expression of cat-2::gfp 4 hours after heat shock, all lost expression after 3 days. Data with a temperature-sensitive allele of ast-1 corroborate sustained ast-1 activity (Supp.Fig.10).

ast-1 is expressed in several neurons 4, including all DA neurons (Fig. 2d) and acts cell-autonomously in DA neurons, as the ast-1 mutant phenotype can be rescued by expression of ast-1 specifically in the DA neurons (Fig. 2e). ast-1 expression persists in DA neurons throughout postembryonic stages, suggesting that ast-1 is not only required to initiate DA terminal cell fate, but to also maintain DA neuron identity, a notion we confirmed through temporally controlled addition and removal of ast-1 gene activity (Fig. 2f; Suppl. Fig.S10).

To address whether ast-1 function is not only necessary for proper DA neuron differentiation, but also sufficient, we ectopically induced ast-1 expression throughout all cell and tissue types at different stages of development (Fig.3). Ectopic induction during embryogenesis leads to a substantial ectopic expression of both dat-1::gfp (Fig.3a-d) and cat-2::gfp (data not shown). The morphology, location and pan-neuronal fate marker expression of these cells suggests that the effects of ast-1 are confined to the nervous system, in which some (20 cells; ∼10% of the embryonic nervous system) but clearly not all cells can be induced to ectopically express both dat-1/DAT and cat-2/TH. Ectopic ast-1 is maximally effective when expressed around the time of neurogenesis (Suppl. Fig. S11). Moreover, ectopic ast-1 expression under control of the DA- and 5-HT-specific bas-1 promoter induces dat-1/DAT expression in 5-HT neurons (Fig. 3e) demonstrating that ast-1 acts autonomously to control DA neuron specification and that 5-HT neurons provide the appropriate cellular context to allow ast-1 to induce DA neuron specification. The related ETS domain transcription factor LIN-1 is not able to induce ectopic DA neuron production when expressed under similar conditions, demonstrating the specificity of AST-1 function.

Fig. 3

Ectopic ast-1 expression can induce DA cell fate

(a) Representative picture of a control embryo after the 3-fold stage. dat-1::gfp expression starts at late three fold stage and can be detected in the six embryonically generated DA neurons.

(b,c) Representative picture of an embryo heat shocked four hours after the two cells stage and analyzed ten hours after the heat shock. dat-1::gfp is ectopically expressed in many cells of the embryo. Scale bar: 20 μm. In the presence of a pan-neuronal marker (rgef-1::rfp)(panel c), the ectopic DA-fate expressing cells can be identified as neurons.

(d,e) Ectopic expression of ast-1 under the control of the ectodermal promoter unc-119 and the DA /5HT-neuron specific promoter bas-1 leads to ectopic expression of dat-1::gfp in additional neurons compared to wild-type worms (red arrowheads). Similar effects were observed in multiple lines (2/2 lines for bas-1 driver; 2/3 lines for unc-119 driver). Scale bar: 100 μm.

To assess whether ETS transcription factor(s) have a similar function in vertebrate DA neuron specification, we analyzed their expression in the DA areas of the brain (Suppl. Table S2). Distinct ETS factors appear to be expressed in distinct types of DA neurons and we focus here on the ETS factor Etv1/ER81, which is expressed in the DA neurons of the olfactory bulb (Suppl. Fig. S12)8,9. Mice lacking Etv1 10 display a dramatic reduction in the number of TH positive cells in their olfactory bulb compared to wild-type siblings whereas other periglomerular interneuron subtypes were not affected or less severely reduced (Fig. 4a-c; Suppl. Fig. S13). This phenotype is not paralleled by increased cell death, nor by a reduction in the overall density of cells in the glomerular layer, nor by a reduction in overall neuron number, nor by proliferation defects (Fig. 4d-f; Suppl. Fig. S14). Moreover, the identity of DA progenitor cells in the lateral ganglionic eminence (LGE), which already express Etv1 11, appears unaffected in Etv1 mutants (Suppl. Fig. S15). Therefore, Etv1 may affect a late stage in olfactory DA neuron differentiation.

Fig. 4

Mouse Etv1/Er81 is necessary for the olfactory bulb dopaminergic neuron specification

(a,b) Coronal section TH immunostaining of a wild type (A) and Etv1 mutant (B) newborn pup (P0) olfactory bulb. Scale bar: 150 μm.

(c) Quantification of TH positive cells in wild type and Etv1 mutants at P0. Etv1 mutants show a significant reduction of the TH positive cells already at this stage (n=3, p-value=0.00009).

(d, e) Coronal section tuj1 immunostaining and DAPI staining of a wild type (d, d’) and Etv1 mutant (e, e’) P0 glomerular layer to label neurons and cell nuclei. Scale bar: 40 μm.

(f) Quantification of DAPI nuclei in wild type and Etv1 mutants at P0. Glomerular layer cell density is similar between wild type and Etv1 mutants (n=3, p-value=0.93).

(g) Overexpression of Etv1 can induce DA differentiation. Dissociated P0 olfactory bulbs were transfected with GFP and PCDNA3.1 (control) or GFP and Etv1 cloned into the PCDNA expression vector, plated and cultured for 4 days. Etv1 overexpression leads to increased number of tyrosine hydroxylase positive cells.

(h) Analysis of the activation of TH promoter by Etv1 in COS cells. The dotted line indicates the level of luciferase activation of the empty luciferase vector observed upon Etv1 transfection. “DA motif” = phylogenetically conserved match to the a VGGAWRNV consensus. n=3 independent experiments for each construct.

Like ast-1, Etv1 appears not only required for DA neuron differentiation but also sufficient, as ectopic expression of Etv1 in olfactory bulb primary cell culture increases the number of cells expressing the DA marker TH (Fig.4g). Etv1 is also able to directly activate the cis-regulatory region of the mouse TH locus in a heterologous context (Fig. 4h). This activation depends on the presence of two phylogenetically conserved DA motifs (Fig. 4h). Like in C. elegans , phylogenetically conserved DA motifs can also be found in the 5’ upstream regulatory region of all four other mouse dopamine pathway genes (Fig. S16). Another indicator for a conserved function of mouse Etv1 and worm AST-1 is that mouse Etv1 is able to rescue the ast-1 mutant phenotype when expressed in transgenic worms (Fig.2e).

In conclusion, we have described here a surprisingly simple regulatory logic for DA specification. Our cis-regulatory analysis in worms reveals that all dopamine pathway genes are co-regulated through a similar cis-regulatory motif and trans-acting factor and this regulatory logic applies to dopaminergic neurons of distinct lineage origin. Our analysis demonstrates that the ETS factor ast-1 is a terminal selector gene for dopaminergic cell fate, akin to other terminal selector genes that control the terminal identity of other neuron types 12-15. Terminal selector genes are transcription factors that directly regulate the “nut-and-bolts” differentiation gene batteries that determine the specific properties of a neuron by binding to simple cis-regulatory motifs shared by members of the terminal gene batteries, termed “terminal selector motifs” (in the case of the DA neurons, the DA motif) 15. As exemplified by AST-1, terminal selector genes are continuously expressed throughout the life of a neuron to ensure that the terminal differentiation state is properly maintained.

The regulatory logic of DA neuron specification appears to be phylogenetically conserved. Vertebrate dopamine pathway genes also contain DA motifs that are required for the activation by a trans-acting factor that is homologous to the C. elegans trans-acting factor. Loss of the trans-acting factor either in worms or mice leads to a loss of the dopaminergic phenotype. Both Etv1 and ast-1 are continuously expressed throughout the postmitotic life of DA neurons, and our analysis in worms indicate that these factors also maintain the terminal identity of DA neurons. The function of vertebrate ETS proteins in DA specification may have been distributed over several different ETS domain transcription factors, as Etv1 is not expressed in other DA neuron population in the brain and as it does not affect the generation of these other types of DA neurons (data not shown). Those other areas express a related ETS factor, Etv5, which may fulfill a similar role as Etv1 in olfactory DA neurons; in support of this notion, Etv5 can also transactivate the TH promoter in a heterologous assay system (Supp.Fig. S17). The logic of distributing an ancestral gene function, observed in an invertebrate species, over several vertebrates paralogs of the ancestral invertebrate ortholog has been noted for other transcription factors as well 16 and appears an important component of driving neuronal diversification processes in more complex brains.

While AST-1 and Etv1 both act as selector genes for DA terminal differentiation, their presence is not strictly sufficient to activate DA genes as both AST-1 and Etv1 are expressed in other neurons apart from DA neurons 4,17. Our ectopic expression experiments also show that AST-1's ability to induce ectopic DA fate is restricted to some cellular and temporal contexts. Classic “master regulators”, like eyeless or MyoD also show similar context-dependencies in their mode of action 18,19. AST-1 and Etv1 function may be actively inhibited in cells “refractory” to AST-1/Etv1 activity. Alternatively, AST-1/Etv1 function may require additional, cell type-specific factors for appropriate function in DA neurons. Such “combinatorial coding” mechanisms are a common theme in neuron type specification 20 and our identification of a conserved role of ETS factors as a central component of such a code is the first important step in decoding the regulatory logic of DA neuron specification. It will be interesting to see whether the additional specificity determinants of ETS factors are also conserved from worms to vertebrates.

METHOD SUMMARY

Transgenic and mutant C.elegans strains

Reporter gene constructs were generated by subcloning into the pPD95.75 backbone vector, which contains the gfp coding sequence and the unc-54 3’ UTR. Mutagenesis and deletions were performed using the Quickchange II XL Site-Directed Mutagenesis Kit (Stratagene). Reporter constructs were injected into otIs181( Is[dat-1::mCherry;ttx-3::mCherry]) to facilitate the identification of the DA cells. DNA was injected at 50ng/μl using rol-6 as injection marker. For every construct, 30 or more animals were scored from at least 2 different transgenic lines. Ets mutant strains were obtained from the CGC. For the heat shock experiments, a transgenic strain of the following genotype was generated: OH7546 [otIs198(hsp16−2::ast-1;hsp16−2::NLS-mCherry;ttx-3::ds-red),vtIs1(dat-1::gfp;rol6)]. Heat-shock induction conditions, as well as a list of transgenic and mutants strains, are provided in the Suppl. Methods.

Analysis of vertebrate Etv1

Standard histological protocols were used to analyze wild-type and Etv1 mutant mouse samples. For ectopic Etv1 expression, olfactory bulbs were dissected from P0-P1 wild-type mice, dissociated and electroporated using the Amaxa Nucleofector System and the mouse neuron kit, following the manufacturer's protocol. For the TH promoter analysis, COS cells were transfected with Lipofectamine (Invitrogen) and luciferase activity was measured using the Luciferase Assay kit (Stratagene) and the ß-galactosidase Enzyme Assay System (Promega). See supplementary methods for full description on histology, cell culture and quantification methods.