A subunit of the mediator complex regulates vertebrate neuronal development

Abstract

The unique profiles of gene expression dictate distinct cellular identity. How these profiles are established during development is not clear. Here we report that the mutant motionless (mot), identified in a genetic screen for mutations that affect neuronal development in zebrafish, displays deficits of monoaminergic neurons and cranial sensory ganglia, whereas expression of the pan-neuronal marker Hu is largely unperturbed; GABAergic and subsets of cranial motor neurons do not appear to be deficient. Positional cloning reveals that mot encodes Med12, a component of the evolutionarily conserved Mediator complex, whose in vivo function is not well understood in vertebrates. mot/med12 transcripts are enriched in the embryonic brain and appear distinct from two other Mediator components Med17 and Med21. Delivery of human med12 RNA into zebrafish restores normality to the mot mutant and, strikingly, leads to premature neuronal differentiation and an increased production of monoaminergic neuronal subtypes in WT. Further investigation reveals that mot/med12 is necessary to regulate, and when overexpressed is capable of increasing, the expression of distinct neuronal determination genes, including zash1a and lim1, and serves as an in vivo cofactor for Sox9 in this process. Together, our analyses reveal a regulatory role of Mot/Med12 in vertebrate neuronal development.

During vertebrate development, pluripotent stem cells respond to spatially localized signals that regulate their cell cycle exit and subsequent differentiation into specialized cell types. The central nervous system contains a large number of different cell types and has been an organ of interest for studying progenitor cell commitment/differentiation and the generation of cellular diversity (1, 2). In addition to spatial control, temporal regulation of neuronal development has been appreciated but is much less understood in the vertebrate nervous system (3, 4). It is widely accepted that progenitor cells need to establish unique profiles of gene expression that dictate their final destiny. Intracellular pathways underlying the establishment of precise gene expression patterns in the developing nervous system are not well understood.

We have undertaken a genetic approach to characterize genes and pathways that control vertebrate neuronal development by using zebrafish as a model system (5–8). Here, we describe the molecular characterization of the motionless (mot) mutant isolated from our genetic screen. The mot mutant embryos have defects in movement and neuronal and cardiovascular development (9). Current analyses reveal that they have normal brain patterning and do not suffer a global deficit of neurons as evidenced by largely unperturbed expression of the pan-neuronal marker Hu. However, the mot mutant exhibits deficits in neuronal subtypes that include monoaminergic (MA) neurons [forebrain dopaminergic and serotonergic (5HT) neurons, hindbrain noradrenergic (NA) and 5HT neurons, and neural-crest derived sympathetic neurons] and cranial sensory ganglia, but not in GABAergic neurons and subsets of cranial motor neurons. Positional cloning discloses that the molecular cause of the mutant phenotype is caused by disruption of the Mediator subunit Med12 (previously also known as Trap230). Overexpression of human Med12 in zebrafish leads to premature neuronal differentiation and/or an increased production of brain dopaminergic, 5HT, and NA neurons. Further investigation reveals that mot/med12 is required and sufficient to increase the expression of certain neuronal determination genes including zash1a and lim1. Moreover, proper expression of these genes requires Sox9, which is capable of increasing their expression in a Mot/Med12-dependent manner. These findings uncover an important role of Mot/Med12 in regulating vertebrate neuronal development and identify Sox9 as its genetic partner in regulating downstream genes lim1 and zash1a.

Results

The mot Mutation Disrupts the Development of Neuronal Subtypes in the Central and Peripheral Nervous Systems.

We previously isolated zebrafish mutants affecting dopaminergic and NA neurons in a genetic screen by using the neurotransmitter synthesis enzyme tyrosine hydroxylase (TH) as a molecular marker. The mot mutant phenotypes were highly penetrant and consistent: the mutant displayed certain morphological abnormalities, including mild cyclopia, brain ventricles that failed to be properly inflated, leading to a smaller appearance of the brain, and cardiovascular defects. Also as previously reported, and except for restricted apoptosis in the telencephalon and lens at later stages, no gross cell death was observed (9). Brain-patterning genes including shh, pax2, pax6, patched, foxa2, and krox20 appeared overall normal (Fig. 1A–F and Fig. 8, which is published as supporting information on the PNAS web site). In addition, the general neuronal marker Hu was largely unperturbed in the mot mutant embryo (Fig. 1 G–J).

Fig. 1.

The mot mutant displays overall normal brain patterning and expression of the pan-neuronal marker Hu. Anterior is to the left, and dorsal is up. (A, C, E, G, and I) WT. (B, D, F, H, and J) mot. (A–F) In situ hybridization shows the expression of shh (A and B), pax2 (C and D), and pax6 (E and F). (G–J) Immunostaining shows the pattern of the pan-neuronal marker Hu. fb, forebrain; hb, hindbrain; mb, midbrain; MHB, midbrain hindbrain boundary. (Magnifications: A, B, and G–J, ×200; C–F, ×100.)

Strikingly, however, the mot mutant embryo displayed deficits in many distinct neuronal subtypes, including forebrain dopaminergic neurons, hindbrain locus coeruleus (LC) NA neurons (Fig. 2A and B and Fig. 9, which is published as supporting information on the PNAS web site) (9), hindbrain 5HT neurons (Fig. 2 C and D), cranial sensory ganglia (Figs. 2 E and F and 9 G–N) (VIIth to Xth ganglia were more severely affected than others), and sympathetic NA neurons (Fig. 2 G and H). These neuronal deficits were detected at the onset of their differentiation (Fig. 2B and data not shown), suggesting that the mot gene is required for the determination rather than survival of these neurons.

Fig. 2.

The mot mutation disrupts the development of distinct neuronal classes in the brain. All are lateral views except G, H, and M–P, which are dorsal views. (A, C, E, G, I, K, M, and O) WT. (B, D, F, H, J, L, N, and P) mot. (A and B) Whole-mount in situ hybridization shows TH⁺ forebrain dopaminergic neurons, hindbrain LC NA neurons, and NA arch-associated CA cells. (C and D) 5HT immunostaining shows forebrain and hindbrain 5HT neurons. (E–H) Immunostaining shows Hu⁺ cranial sensory ganglia (E and F) and TH⁺ sympathetic ganglia (G and H). (I–L) In situ hybridization shows gad67⁺ GABAergic neurons. (M and N) In situ hybridization shows phox2a⁺ ocular and trochlear and hindbrain motor nuclei. (O and P) In situ hybridization with islet-1 shows hindbrain cranial motor neurons. AAC, arch-associated CA cells; C, cranial sensory neurons; OT, ocular and trochlear motor neurons; Sym, sympathetic ganglia. (Magnifications: ×400.)

In contrast to deficits of these neuronal classes, brain GABAergic neurons do not appeared to be deficient and perhaps are increased in some locations (Fig. 2 I–L). Also, ocular and trochlear motor nuclei in the mid/hindbrain region appeared increased (Fig. 2 M and N), as evidenced by expression of the paired homeodomain containing gene phox2a (6). In the hindbrain, whereas laterally localized motor nuclei were defective, medially located ones appeared increased (Fig. 2 M and N). Staining with islet-1 confirmed that whereas certain lateral groups were defective, medial ones were found expanded (Fig. 2 O and P). These defects are consistent with either cell fate changes or migration problems. Together, these analyses indicate that the mot mutant does not suffer an early patterning defect and is capable of expressing general neuronal markers, but displays abnormalities in the proper development of neuronal subtypes.

Positional Cloning Identifies mot as Zebrafish Med12, a Subunit of the Mediator Complex.

To identify the mot gene, we mapped the mot mutation to LG 14, between two microsatellite markers Z9017 and Z7495 that are ≈0.1 cM apart (Fig. 3A). Through searches of zebrafish databases and chromosomal walking, we constructed a contig of BAC clones that span the mot region (Fig. 3A). We carried out further mapping by sequencing BAC ends and identifying additional polymorphic markers. After sequencing candidate genes that were predicted in the final mapped region, we found that both the cDNA and genomic DNA from mot mutant embryos carried a deletion of two nucleotides at position 109 of a predicted transcript, resulting in a frame shift and premature truncation of the gene product homologous to Med12 (also known as Trap230) (Fig. 3 B and C). Med12 is a subunit of the Mediator complex identified through biochemical purification in mammalian cells (10). Structural comparison between the predicted Mot protein and the human Med12 revealed a high degree of conservation throughout the proteins (overall amino acid identity of 81%), including the leucine-rich domain (83% amino acid identity), the leucine- and serine-rich domain (83% amino acid identity), the proline-, glutamine-, and lucine-rich domain (53% amino acid identity), and the glutamine-rich domain (Fig. 3B). Together, these analyses suggest that the mot mutation disrupts the zebrafish homologue of Med12.

Fig. 3.

Positional cloning identifies the mot gene as zebrafish med12. (A) The genetic and physical map of the mot locus on zebrafish LG 14 is shown. (B) Schematic drawing shows the structure of the predicted WT mot/med12 gene product and the truncated gene product in the mot^m807 mutant. L, leucine-rich domain; LS, leucine- and serine-rich domain; PQL, proline-, glutamine-, and lucine-rich domain; OPA, glutamine-rich domain. (C) DNA sequence chromatograms show the molecular nature of the mot^m807 mutation.

To further verify this conclusion, we carried out morpholino oligonucleotide (MO) injection experiments. WT embryos injected with a splicing MO targeting med12 gene displayed both morphological (93%, n = 80) and dopaminergic neuronal (95%, n = 40) deficits resembling those of the mot mutant embryo, whereas WT injected with a control MO appeared normal (n = 50) (Fig. 10 A–I, which is published as supporting information on the PNAS web site). Therefore, injection of the MO targeting med12 gene phenocopied the mot mutant defects.

We next determined whether the mot mutant defects might be rescued by injecting med12 mRNA. Because the predicted mot transcript is ≈7 kb long, we had not been able to recover a full-length cDNA clone corresponding to the zebrafish mot/med12. To this end, we made in vitro-transcribed human med12 mRNA and the control β-gal mRNA and separately injected them into the mot mutant embryos. Injection of human med12 mRNA (54%, n = 70) (Fig. 10 L and O) but not the β-gal mRNA (0%, n = 45) (Fig. 10 K and N) readily rescued both morphological and dopaminergic neuronal defects of the mot mutant: the β-gal mRNA-injected mot mutant possessed ≈2 (±1) weakly TH⁺ cells (n = 30), whereas med12 mRNA-injected mutant embryos had an average of 8 (±2) TH⁺ cells that showed a WT level of TH immunoreactivity (n = 30). Taken together, these analyses demonstrate that the mot gene encodes a functional orthologue of the mammalian Med12.

Expression of mot/med12 and Two Other Mediator Subunit-Encoding Genes med17 and med21 in the Developing Embryo.

To determine the spatial and temporal expression pattern of mot/med12, we carried out whole-mount in situ hybridization. mot/med12 was maternally expressed (Fig. 4A and B). During somitogenesis stages, mot/med12 was broadly expressed in the brain but showed enrichment in the brain ventricles (Fig. 4 C and E). Such staining pattern was dramatically reduced in the mot mutant embryo (Fig. 4 D and F), possibly because of nonsense-mediated RNA decay triggered by the early encounter of a stop codon. This loss of expression in the mutant also validates that the pattern detected in WT reflects the true expression profile of mot/med12.

Fig. 4.

Expression of mot/med12, med17, and med21 is shown. (A and B) Maternal mot/med12 expression. (C and D) mot/med12 expression in 24-hpf WT and the mot mutant embryo. (E and F) Higher magnification view of the ≈24-hpf embryonic brain showing mot/med12 expression (Inset is sectioned image). (G and H) mot/med12 expression in 48 hpf embryos shows enrichment in clusters of cells dorsal and posterior to the eyes (black arrows), and such staining is significantly reduced in the mot mutant. White arrows indicate background staining. (I–N) Expression of med17 (I, K, and M) and med21 (J, L, and N) in 24- and 48-hpf embryos. (Magnifications: ×200.)

By 48 hours postfertilization (hpf), mot/med12 expression appeared decreased throughout WT embryos, except that enrichment was found in clusters of cells dorsal and posterior to the eyes (Fig. 4G). Such expression was dramatically reduced in the mot mutant (Fig. 4H), suggesting that it is indeed specific for mot/med12. The identity of these cells should be determined through future analyses with additional markers and will not be discussed further in this study. These results demonstrate that mot/med12 displays an enriched expression pattern in the developing embryonic brain, especially in cells lining the brain ventricles.

To determine whether the enriched pattern of med12 expression is unique to this subunit or whether it represents distribution of the entire Mediator complex, we examined the expression patterns of med17/trap80 and med21/trap19, both of which are thought to encode the core components of the Mediator complex. Neither med17 nor med21 transcripts showed significant enrichment in the brain ventricles (compare Fig. 4 I–J with C and compare Fig. 4 K and L with E). At 48 hpf, neither med17 nor med21 transcripts were enriched in the same clusters of cells that showed high-level expression of mot/med12 (compare Fig. 4 M and N with G). Taken together, these analyses suggest that components of the Mediator complex are present at varying concentrations in various cell types of developing zebrafish.

Overexpression of mot/med12 Leads to Premature Differentiation and an Increased Production of MA Neuronal Subtypes.

Given the requirement of mot/med12 for the development of vertebrate neuronal subtypes, we next determined the impact of overexpressing mot/med12 on neuronal development. We examined MA (dopaminergic, NA, and 5HT) neurons, which were defective in the mot mutant, and GABAergic neurons, which were not deficient in the mot mutant, and the pan-neuronal marker Hu. Injection of in vitro-transcribed human med-12 mRNA into WT embryos led to premature neuronal differentiation as evidenced by the detection of dopaminergic neurons (94% of n = 32) at a stage when they were never detectable in the control embryos (n = 30, none had dopaminergic neurons) (Fig. 5A and B). In WT, LC NA neurons are derived from dorsally located progenitor cells, which do not initiate differentiation until they reach their final ventrolateral location (6) (Fig. 5C). In a portion of med12 RNA-overexpressed embryos, TH⁺ cells were detected at a more dorsal location, suggesting a premature differentiation of LC progenitor cells (Fig. 5D, 17% of n = 18). These observations suggest that overexpression of med12 can lead to premature neuronal differentiation.

Fig. 5.

Overexpression of mot/med12 in WT embryos is shown. All are lateral views. (A, C, E, G, I, and K) WT injected with β-gal RNA. (B, D, F, H, J, and L) WT injected with med12 RNA. (A–F) TH immunostaining shows that TH⁺ dopaminergic neurons (B) and LC NA neurons (D) are detected precociously in med12 RNA-injected embryos; dopaminergic (F) and LC NA (D) are also increased in med12 RNA-injected embryos. (G and H) 5HT neuron increase in med12 RNA-injected embryos. (I–L) Hu labeling (I and J) and GAD67 in situ hybridization (K and L) show comparable patterns between control and the med12 RNA-injected embryos. (Magnifications: A–H, ×200; I–L, ×100.)

Moreover, we detected a significant increase of forebrain dopaminergic neurons in med-12 RNA-injected embryos (Fig. 5 E and F) (62% embryos showed significantly more dopaminergic neurons compared with β-gal-injected embryos, n = 92). A quantitative analysis of the number of dopaminergic neurons revealed an average number of 14 dopaminergic neurons (±2, n = 57) in med-12 RNA-injected embryos versus an average number of 8 (±2, n = 40) dopaminergic neurons in β-gal-injected embryos. Overexpression of med12 RNA also increased the production of LC NA neurons (Fig. 5D; 78% of n = 18) and 5HT neurons (Fig. 5H; 85% of n = 20). However, overexpression of med12 did not appear to significantly increase general neuronal production as evidenced by the apparently normal Hu staining (Fig. 5 I–J), nor did it increase GABAergic neurons significantly (Fig. 5 K and L). These analyses indicate that Mot/Med12 can promote the differentiation of brain MA neurons.

mot/med12 Is Necessary and Sufficient to Regulate the Expression of Distinct Neuronal Determination Genes.

The loss- and gain-of-function studies suggest that mot/med12 plays an important role in neuronal differentiation. To identify potential downstream target genes of Mot/Med12, we asked whether it controls the expression of neuronal determination genes, which are expressed in progenitor cells and regulate their commitment and differentiation. We found that zash1a (Fig. 6A and B) and lim1 (Fig. 6 C and D) expression was severely defective, whereas dlx2 expression (Fig. 6 E and F) remained largely normal in the mot mutant. zash1a encodes a basic helix–loop–helix transcription regulator (11) and is the zebrafish homologue of mammalian mash1, which is important for the development of central NA and 5HT neurons (12). lim1 encodes a LIM homeodomain (HD) transcription regulator (13) and is expressed in the forebrain and hindbrain progenitor cells that are in close proximity to forebrain and hindbrain MA neurons. dlx2 encodes the distaless-like HD transcription regulator (14) and is shown to be important for the development of forebrain GABAergic neurons (15, 16). The largely normal dlx2 expression was consistent with the lack of apparent GABAergic neuronal deficiency in the mot mutant.

Fig. 6.

mot/med12 regulates the expression of distinct neuronal determination genes. (A–F) In situ hybridization shows the expression of zash1a, lim1, and dlx2 in the brain of WT (A, C, and E) and the mot mutant (B, D, and F) embryos. (G–L) In situ hybridization shows the expression of zash1a, lim1, and dlx2 in WT embryos injected with β-gal RNA (G, I, and K) and human med12 mRNA (H, J, and L). (Magnifications: ×100.)

We next asked whether overexpression of mot/med12 is sufficient to up-regulate zash1a and lim1, which could imply a direct regulation of these genes by mot/med12. Injection of mot/med12 RNA into the WT led to a striking increase of zash1a expression (Fig. 6 G and H, 95%, n = 21), and moderate increase of lim1 expression (Fig. 6 I and J, 84%, n = 25), but did not significantly affect the expression of dlx2 (Fig. 6 K and L, n = 19). These analyses suggest that mot/med12 regulates the proper expression of distinct neuronal determination genes and identify lim1 and zash1a as its potential direct downstream targets.

Mot/Med12 Serves as an in Vivo Cofactor for Sox9 in Regulating the Expression of lim1 and zash1a.

We next asked which transcription regulator Mot/Med12 might interact with in regulating the expression of lim1 and zash1a. A biochemical interaction was previously identified between human Med12 and Sox9 (17), a high mobility group-type DNA-binding transcription regulator involved in chondrocyte differentiation and male sex determination (18, 19). In zebrafish, two sox9 genes have been identified and shown to be involved in craniofacial and pectoral fin development (20). Interestingly, both sox9 genes are expressed in the developing CNS (21), albeit their roles in brain development have not been previously elucidated.

We found that both lim1 and zash1a expression are reduced in the sox9a⁻sox9b⁻ double mutant (Fig. 7A–D). Moreover, injection of sox9a RNA into the WT embryos led to a significant up-regulation of lim1 (Fig. 7 E and F, 78%, n = 44) and zash1a (Fig. 7 G and H, 96%, n = 39) expression. These observations suggest that Sox9 is a critical regulator of lim1 and zash1a expression in the developing zebrafish brain.

Fig. 7.

Sox9 regulates lim1 and zash1a expression in a mot/med12-dependent manner. All are lateral views of the embryonic brain. (A–D) lim1 and zash1a expression in WT (A and C) and sox9a⁻sox9b⁻ double mutant (B and D) embryos. (E–H) lim1 and zash1a expression in WT embryos injected with β-gal RNA (E and G) and sox9a mRNA (F and H). (I–L) lim1 and zash1a expression in WT (I and K) and mot mutant (J and L) embryos injected with sox9a mRNA. (Magnifications: A–H, 200×; I–L, ×100.)

Given the dual requirement of Sox9 and Mot/Med12 in regulating the expression of lim1 and zash1a, and their known biochemical interaction, we next asked whether a genetic interaction exists between these two factors in vivo. By injecting the sox9a RNA into embryos derived from mot heterozygous matings, we found that although sox9a is capable of increasing the expression of lim1 (Fig. 7I, 80%, n = 41) and zash1a (Fig. 7K, 98%, n = 38) in the WT siblings, it failed to do so in the mot mutant embryos (Fig. 7J, 100%, n = 9; Fig. 7K, 100%, n = 11). Taken together, these analyses suggest that Sox9 regulates neuronal determination genes lim1 and zash1a in a Mot/Med12-dependent manner.

Discussion

We have shown that the zebrafish mutant mot exhibits deficits of certain neuronal subtypes, including MA neurons and cranial sensory ganglia but not GABAergic neurons, and show rather normal expression of the Pan-neuronal marker Hu. Through positional cloning, we demonstrate that the mutant phenotype is caused by the disruption of a gene encoding the Mediator subunit Med12. We further show that overexpression of mot/med12 is capable of inducing premature neuronal differentiation and an increased production of brain dopaminergic, NA, and 5HT neurons. Finally, we identify the neuronal determination genes lim1 and zash1a as Mot/Mediz's potential downstream targets and demonstrate a genetic dependence of sox9 on mot/med12 in regulating lim1 and zash1a. These findings uncover a regulatory role of the Mediator complex in vertebrate neuronal development.

The Phenotypes of the mot Mutant.

The mot mutant does not have a global deficit in neuronal development, because regional patterning of the nervous system is apparently normal. Moreover, expression of the pan-neuronal marker Hu is largely unperturbed; whereas certain neuronal subtypes that include MA (dopaminergic, NA, and 5HT) and cranial sensory neurons are defective, others such as GABAergic and subsets of cranial motor neurons are not. Although other neuronal types in addition to the ones described in the study may be affected, it is clear that the mot mutant does not affect all neurons. These observations suggest that the mot/med12 gene plays a regulatory role in vertebrate neuronal development. Consistent with this notion, it is interesting to mention that Caenorhabditis elegans (22–25) and Drosophila (26, 27) mutants in med12 also display specific phenotypes, and human polymorphisms in med12/hopa correlate with distinct brain diseases such as schizophrenia and hypothyroidism (28).

Although the focus of our current study is on vertebrate neuronal development, it is worth pointing out that the mot mutant has additional defects that include cardiovascular abnormalities. Bradycardia and blood accumulation near the heart was observed in 2-day-old mot mutant embryos (9), and other organ defects are similar to what have been reported for other alleles of med12 (29, 30). Given the pattern of neuronal defects observed in this study, we speculate that mot may also play a regulatory role in the development of heart and possibly other major vertebrate organs.

Regulation of Neuronal Development by Mot/Med12.

The development of multipotent stem cells into many thousands of specialized neuronal types in the vertebrate nervous system involves intricate regulation by extrinsic signals and intrinsic determinants. Although cascades of region-specific DNA-binding transcription factors are appreciated in the determination of specific neuronal lineages, the identification of a Mediator subunit in regulating this process is somewhat unexpected. Because the expression of mot/med12 is not restricted to specific classes of neuronal progenitors, it is less obvious how it exerts differential effects on neuronal development. Several possible scenarios may be considered: (i) The mot mutant phenotype may be caused by a partial loss of function of Med12. This scenario is unlikely, because our molecular genetic analyses indicate that the deficits in the mot mutant embryo arise from severe premature truncation of the protein, and therefore, are likely to render it nonfunctional. The fact that the phenotype of morpholino-injected embryos closely resembles that of the mot mutant is consistent with this notion. (ii) The gene- and neuron-selective effects may be caused by differential persistence of maternal RNA in different cells. Because of the instability of zygotic mutant mot/med12 RNA, we are able to essentially examine the maternal RNA distribution in the mot mutant embryo (Fig. 4F); no obvious spatially and temporally distinct pattern was detected, suggesting that differential persistence of maternal RNA is also an unlikely explanation. (iii) The gene- and neuron-selective effects may be caused by compensation by other Mediator subunits. The Mediator complex is indeed composed of many (≈30) subunits and is a transcriptional coactivator complex that was described in yeast (31–33) and mammalian cells (34). However, so far only one med12-like gene has been found in the zebrafish genome, and individual Mediator subunits appear to be quite distinct and show little sequence homologies. Thus, compensation by other Mediator subunits is also an unlikely explanation for the gene- and neuron-selective effects of Mot/Med12.

Our findings that the expression of distinct neuronal determination genes such as zash1a and lim1 is defective, whereas that of dlx2 appears largely unperturbed, suggest that despite its wide expression, mot/med12 is capable of conferring specific regulation on gene expression. Specificity of transcriptional regulation has been demonstrated for several other Mediator subunits. For instance, Med1 (also known as Trap220) is required for nuclear receptor proliferator-activated receptor γ2-stimulated adipogenesis but is dispensable for MyoD-mediated myogenesis (35). Likewise, Med15 (also known as Arc-105) is required for TGF-β/Nodal signaling but not for BMP/Smad1 signaling (36), and Med23 (also known as Sur2) is selectively required for transcription factors activated in MAP kinase pathways (37). Therefore, Mot/Med12 may exert its selective effects by interacting with spatially and temporally distinct transcription factors. The observed genetic interaction between Sox9 and Mot/Med12 in regulating lim1 and zash1a expression (this study), together with a recent report (30), and the demonstrated biochemical interaction between these two factors (17), provides strong evidence that Mot/Med12 is a cofactor for Sox9 in vivo. Despite this, it is also worth pointing out that Mot/Med12 may serve as a cofactor for additional transcription regulators, because dopaminergic neurons, which are defective in the mot mutant, remain largely normal in the sox9a⁻sox9b⁻ double mutant (data not shown). Given a recent reported biochemical interaction between Med12 and β-catenin (38), it is of great interest to determine the transcription regulators that interact with Mot/Med12 in regulating dopaminergic neuronal development.

Med12 belongs to a submodule that has been implicated in transcriptional repression in yeast (39) and transcriptional activation in Drosophila (40). Given that Med12 is found to interact with Sox9's transcriptional activation domain (17), and that both factors are required for the activation of lim1 and zash1a, we favor the hypothesis that Mot/Med12 acts as a coactivator for Sox9 in regulating lim1 and zash1a expression.

The Expression of mot/med12 and Other Mediator Subunits Suggest Heterogeneity and Functional Specificity of the Mediator Complex in Vivo.

The Mediator complex is thought to act as a “bridge” between DNA-binding initiation factors and RNA polymerase II in vitro. However, the complexity of the Mediator implies a perhaps more sophisticated role than being a simple bridge in vivo. Recent biochemical characterizations of Med1/Trap220 suggest that heterogeneity and functional specificity might exist for the Mediator complex (41). Our finding of enriched expression of mot/med12 in brain ventricle cells and specific cell types during later development, which do not show enrichment of two other mediator subunits examined, and the recently reported expression of Med24/Trap100 (42), provide in vivo evidence that components of the Mediator complex are present at varying concentrations in various cell types. Hence, the Mediator complex is likely to possess extensive composition heterogeneity and functional specificity and may serve as a module for integrating complex transcriptional regulation in vivo.

Materials and Methods

Fish Stocks and Maintenance.

Fish breeding and maintaining were performed as described (9). Embryos were raised at 28.5°C and staged according to ref. 43. The mot mutation used in the study was isolated in a genetic screen with the allele number m807 (9).

In Situ Hybridization and Immunostaining.

RNA probes were synthesized from linearized templates using RNA labeling reagents (Roche, Indianapolis, IN). TH and 5HT antibodies were purchased from Chemicon (Temecula, CA) and Immunostar (Hudson, WI), respectively. The procedures of in situ hybridization and antibody staining were performed as described (6).

Genetic Mapping and Positional Cloning.

AB/EK female fish carrying the mot^m807 mutation were crossed to WT WIK male fish, and F₁ progeny were raised to adulthood. Genomic DNA was extracted from pools of ≈40 F₂ WT sibling and ≈40 mot mutant embryos, and PCRs were performed with microsatellite marker primers, the sequences of which were obtained from the Massachusetts General Hospital zebrafish database (http://zebrafish.mgh.harvard.edu). Once a pair of closely linked primers that flank the mot locus were identified, they were used to test ≈5,000 individual mutant embryos. BAC clones in the mapped region were identified from the zebrafish genomic database located at the Zebrafish Information Network (http://zfin.org). Specific PCR primers were designed according to the BAC end sequences and used to further narrow down the mot locus. A partial zebrafish med12 cDNA clone was obtained by RT-PCR and cloning, and 5′-RACE (BD Biosciences, Franklin Lakes, NJ) was carried out to identify the missing 5′ end of the cDNA.

Sequence Analysis and Mutation Detection.

Sequence analysis was done by using the GCG program. For mutation detection, gene-specific primers were used to amplify genomic DNA from pools of (≈5) mot mutant and WT sibling embryos. PCR products from mutant and WT sibling embryos (three independent sets) were directly sequenced by using automated cycle sequencers (Applied Biosystems, Foster City, CA). The identified molecular lesion from genomic DNA sequencing was also confirmed by direct sequencing cDNA products from mutant and WT sibling embryos.

Microinjection.

Capped sense RNA was synthesized with a mMESSAGE mMACHINE kit (Ambion, Austin, TX) from linearized pCS2 plasmids encoding either a full-length med12 or β-gal gene. In vitro-transcribed RNA (0.5–1 nl at 100–500 ng/μl) was microinjected into one- to eight-cell stage embryos. MOs were obtained from Gene Tools (Philomath, OR). A MO targeted to the fourth intro/exon boundary (sequence: acacactgacctatacaggggtcga) and a standard control MO (purchased from Gene Tools) were injected at a concentration of 0.5–1.5 mM.

Abbreviations

mot

motionless

MA

monoaminergic

5HT

serotonergic

NA

noradrenergic

TH

tyrosine hydroxylase

MO

morpholino oligonucleotide

hpf

hours postfertilization

LC

locus coeruleus.