phospholipase C, beta 3 is required for Endothelin1 regulation of pharyngeal arch patterning in zebrafish

Abstract

Genetic and pharmacological studies demonstrate that Endothelin1 (Edn1) is a key signaling molecule for patterning the facial skeleton in fish, chicks, and mice. When Edn1 function is reduced early in development the ventral lower jaw and supporting structures are reduced in size and often fused to their dorsal upper jaw counterparts. We show that schmerle (she) encodes a zebrafish ortholog of Phospholipase C, beta 3 (Plcβ3) required in cranial neural crest cells for Edn1 regulation of pharyngeal arch patterning. Sequencing and cosegregation demonstrates that two independent she (plcβ3) alleles have missense mutations in conserved residues within the catalytic domains of Plcβ3. Homozygous plcβ3 mutants are phenotypically similar to edn1 mutants and exhibit a strong arch expression defect in Edn1-dependent Distalless (Dlx) genes as well as expression defects in several Edn1-dependent intermediate and ventral arch domain transcription factors. plcβ3 also genetically interacts with edn1, supporting a model in which Edn1 signals through a G protein-coupled receptor to activate Plcβ3. Mild skeletal defects occur in plcβ3 heterozygotes, showing the plcβ3 mutations are partially dominant. Through a morpholino-mediated deletion in the N-terminal PH domain of Plcβ3, we observe a partial rescue of facial skeletal defects in homozygous plcβ3 mutants, supporting a hypothesis that an intact PH domain is necessary for the partial dominance we observe. In addition, through mosaic analyses, we show that wild-type neural crest cells can efficiently rescue facial skeletal defects in homozygous plcβ3 mutants, demonstrating that Plcβ3 function is required in neural crest cells and not other cell types to pattern the facial skeleton.

Introduction

Genetic analyses in mice support the hypothesis that during embryonic development the ligand Endothelin1 (Edn1) signals through a receptor [Endothelin receptor type A (Ednra)] that is coupled to G proteins of the G_q/G₁₁ family (Clouthier et al., 2000; Dettlaff-Swiercz et al., 2005; Ivey et al., 2003; Kurihara et al., 1994). Mice carrying mutations in Edn1, Ednra, and G_q/G₁₁ have similar facial skeletal defects in which the ventral lower jaw cartilages are severely reduced in size and are fused to their dorsal upper jaw counterparts. Zebrafish with mutations in furinA (an activator of the Edn1 ligand) and edn1 (Piotrowski et al., 1996) also have facial skeletal defects resembling those of the mouse Edn1 pathway mutants, demonstrating that the Edn1 signaling pathway functions broadly within jawed vertebrates to give ventral character to the facial skeleton (Miller et al., 2000; Walker et al., 2006). In this paper we expand the number of Edn1 pathway mutants so far known in either mouse or zebrafish by showing that the craniofacial gene schmerle encodes a zebrafish ortholog of Plcβ3, that plcβ3 mutants have facial skeletal defects similar to edn1 mutants, and that plcβ3 and edn1 genetically interact. We will refer to schmerle (she) as plcβ3 for the remainder of the paper.

Phospholipase C enzymes are effectors of signal transduction pathways coupling an agonist-stimulated cell surface receptor to the intracellular production of the secondary messengers, IP3 and DAG (Rebecchi and Pentyala, 2000; Rhee, 2001). These messengers then promote the activation of protein kinase C and the release of Ca²⁺ from intracellular stores, inducing a wide variety of cellular functions. Plcβ3 is member of the beta class of Phospholipase C enzymes containing four family members, Plcβ1–4 (Rebecchi and Pentyala, 2000; Rhee, 2001). Plcβs are modular proteins containing an N-terminal PH (pleckstrin homology) domain, EF-hand domains, catalytic X and Y domains, a C2 domain, and a C-terminal myosin tail. The PH domain binds membrane phosphoinositides and promotes targeting of Plcβ enzymes to the plasma membrane. The catalytic X and Y domains are both required for the hydrolysis of PIP2 to IP3 and DAG. The myosin tail domain is unique to Plcβ type enzymes. All four Plcβ family members are can be activated by G proteins of G_q/G₁₁ family, and all but Plcβ4 are also activated by Gβγ (Lee et al., 1994; Smrcka and Sternweis, 1993).

Mice bearing mutations in Plcβ1, Plcβ2, and Plcβ4 family members are homozygous viable but exhibit specific defects (Chakrabarti et al., 2003; Li et al., 2000). Plcβ1 mutants die due to epileptic-like seizures, Plcβ2 mutants have enhanced responses to chemoattractants, and Plcβ4 mutants have defects in motor coordination and vision (Hashimoto et al., 2001). Two independent Plcβ3 mouse mutants have markedly different phenotypes (Wang et al., 1998; Xie et al., 1999). Xie et al. (1999) report that Plcβ3 deficient mice are homozygous viable with skin tumors and enhanced sensitivity to morphine, while Wang et al. (1998) report that Plcβ3 deficient mutants are early embryonic lethal. These discrepancies in Plcβ3 mutant mouse phenotypes may be due to differences in the deletion constructs used.

We report that independent zebrafish mutations in conserved X and Y catalytic domain residues of Plcβ3 yield facial skeletal phenotypes similar to edn1 pathway mutants (Miller et al., 2000). Heterozygous plcβ3 mutants have mild facial skeletal phenotypes resembling a partial loss of Edn1 function, while homozygous plcβ3 mutants have severe facial skeletal phenotypes resembling a strong loss of Edn1 function (Miller and Kimmel, 2001). We provide evidence that the plcβ3 catalytic domain mutations act in a dominant negative manner, affecting the function not only of Plcβ3 but of other Plcβ family members as well. We further show through mosaic analyses that Plcβ3 function is autonomously required in neural crest cells specifically to pattern the facial skeleton.

Materials and methods

Fish stocks and maintenance

Fish were raised under standard conditions and occasionally with the addition of 0.0015% PTU (1-phenyl 2-thiourea) to inhibit melanogenesis (Westerfield, 1993). Stages are given in hours postfertilization (hpf) at 28.5°C (Kimmel et al., 1995). plcβ3^tg203e and plcβ3^th210 alleles, which are homozygous lethal, were generously provided by Drs. Tatjana Piotrowski and Christiane Nusslein-Volhard. The sucker^tf216b (endothelin1) mutants have been previously described (Miller et al., 2000). We obtained homozygous mutant embryos from natural matings of heterozygous carriers maintained on an inbred AB genetic background. plcβ3^th210 was used for all phenotypic analyses unless noted otherwise. fli1-GFP albino transgenic fish, TG(fli1:EGFP)^y1; alb^b4, have been previously described (Lawson and Weinstein, 2002).

Tissue labeling procedures

Alcian Green was used to stain the cartilage of 4–6 days postfertilization (dpf) fixed larvae. Facial cartilages were dissected out and prepared as flat mounts (Kimmel et al., 1998). Double Alcian Blue (cartilage) and Alizarin Red (bone) staining was done on 4–6 dpf fixed larvae using a modified “acid-free” protocol (Walker and Kimmel, 2006). Wholemount RNA in situ hybridizations were performed using digoxigenin-labeled riboprobes (Miller et al., 2000). plcβ3 probe was made with a 2 kb c-terminal cDNA clone. References for other probes are: dlx5a, dlx6a, barx1 (Walker et al., 2006), dlx2a, dlx3b, gsc, and hand2 (Miller et al., 2000), bapx1 (Miller et al., 2003), runx2b (Flores et al., 2004), islet1 (Appel et al., 1995) and sox9a (Yan et al., 2002).

Mapping, sequence, and co-segregation analysis of plcβ3 Alleles

We first mapped the schmerle allele, she^tg203e to chromosome 7 on a hybrid TF/AB genetic background using bulked segregant analysis (Knapik et al., 1996). Individual diploid embryos were sorted into mutant and wild-type classes based on an open-mouth phenotype. The heads of these embryos were stained with Alcian Green to confirm segregation of the mouth phenotypes with jaw skeletal defects, and DNA was prepared from the tails and amplified using primers flanking simple sequence-length polymorphisms. Finer mapping placed she^tg203e between the microsatellite markers Z7958 and Z8540 using a mapping panel containing 67 mutant and 29 wild-type diploid embryos. cDNAs were prepared from pools of mutant and wild-type sibling embryos by RT-PCR. Briefly, total RNA was isolated with Trizol (BRL) and primed for first strand cDNA synthesis using Superscript II reverse transcriptase (BRL) and oligo dT primer. Using sequence from the genomic clones Al772136 and BX511067 we designed primers to obtain overlapping cDNAs covering the full-length coding region of plcβ3 as follows: RT-PCR using external primers 5′CCGTTGTTACACTGAAGGT3′/5′GTTTAGTGCCACCATCTG3′ followed by nested primers 5′GAAACTTCCAACCGAGAAG3′/5′CACGTTCCAGAAAAGCTGTG3′ yielded an n-terminal 2.0 kb plcβ3 cDNA. RT-PCR using external primers 5′GCAAATGAGTCGCATTTACC3′/5′CGTACAACTTCTAGTAATAC3′ followed by nested primers 5′CAAAGGGACTCGTGTGGAC3′/5′GAGAATAGAAATCGATTTGAAG3′ yielded a c-terminal 2.0 kb cDNA. These PCR products were sequenced directly using an ABI automated sequencer.

Morpholino antisense oligonucleotide injections

Translation blocking and splice site blockings morpholinos (MOs) were purchased from Gene Tools, Inc.

plcβ3 splice site MO: 5′TTGTCGTGGTTACCTTGCAATAGCC3′ plcβ3 translation MO: 5′CATGGCTGCTGAATCGACGGGTGG3′ (Sequence complementary to the start codon is underlined).

Roughly 2 nl of MOs diluted to 0.5–10 mg/ml in 0.2M KCl and 0.2% Phenol Red was pressure-injected into the yolk of 1–4 cell zebrafish embryos. For Table 1, the following MO amounts were used: WT + splice MO, 10 ng; WT + translation MO, 6 ng; WT + splice/translation MO, 10 ng and 3 ng respectively; plcβ3^−/− + splice MO, 6 ng.

Table 1.

Penetrance of skeletal phenotypes in plcβ3 mutants

A. Cartilage Phenotypes
Class	Number Scored	Joint Loss		Severe Ventral Cartilage Loss
		A1	A2	A1	A2	A3	A4	A5	A6	A7
−/−	164	100%	100%	62%	89%	80%	26%	0%	0%	0%
+/−	148	26%	0%	0%	0%	0%	0%	0%	0%	0%
+/+	70	0%	0%	0%	0%	0%	0%	0%	0%	0%
Class	Number Scored	Number ectopic cartilage nodules/embryo
		0	1	2	3	3+
−/−	164	1%	3%	8%	19%	68%
+/−	148	28%	49%	17%	5%	1%
+/+	70	0%	0%	0%	0%	0%
B. Bone Phenotypes
Class	Number Scored	Bone Loss		Bone Gain		Bone Fusion
		OP	BR	OP	BR	OP/BR
−/−	164	70%	99%	18%	0%	1%
+/−	148	0%	0%	8%	0%	3%
+/+	70	0%	0%	0%	0%	0%

At 6 dpf larvae were fixed and stained for both cartilage and bone.

‘Number Scored’ refers to embryos scored. Severe ventral cartilage loss is defined as cartilages that are less than half the size of its wildtype counterpart. For the class ‘+/−‘ only larvae with phenotypes are included in table.

Tissue transplantation and confocal imaging

Tissue transplants were as described (Crump et al., 2004a; Crump et al., 2004b). An ‘Alexa568’ mixture of 2% Alexa Fluor 568 dextran and 3% lysine-fixable biotin dextran (10,000 MW, Molecular Probes) was injected into the yolk of donor embryos at 1–4 cell stage. For neural crest transplants donor tissue was taken from the animal cap at shield stage (6 hpf) and moved to shield stage host embryos to a position approximately 90° from the shield and 70° from the animal pole. For ectoderm transplants animal cap donor transplants were moved to shield stage host embryos to a position approximately 120° from the shield and 40° from the animal pole. For endoderm transplants donor embryos were additionally injected with TAR*RNA and margin donor tissue moved to 40% epiboly hosts (4 hpf) to margin position. All transplants were unilateral. Host embryos were screened at 30–36 hpf and 6 dpf using a Leica MZFLIII fluorescence stereomicroscope and animals with greater than a quarter donor tissue contributing to neural crest, endoderm, or ectoderm were selected for analysis. Animals were imaged using a Zeiss LSM5 Pascal confocal fluorescence microscope and subsequently fixed and doubly stained for cartilage (Alcian Blue) and bone (Alizarin Red) at 6 dpf.

GenBank Accession numbers

plcβ3 xxxx

Results

plcβ3 mutants have defects in the intermediate and ventral domain elements of the pharyngeal skeleton

Two schmerle (phospholipase C, beta 3) alleles, which we shorten to plcβ3 (see evidence presented below), were isolated in a large-scale mutagenesis screen and placed into a class with mutants at three other loci [sucker (edn1), sturgeon (furinA), and hoover] which disrupt dorso-ventral patterning of the anterior pharyngeal arches (Miller et al., 2000; Piotrowski et al., 1996). In plcβ3 homozygotes the dorsal and ventral cartilages of the anterior arches are fused at an intermediate position corresponding to joint regions, and ventral cartilages are reduced in length in arches 1–4, but not in arches 5–7 (Fig. 1E,H, Table 1). In particular, the ventral cartilage (ceratohyal) of the second arch is severely reduced in homozygotes, with nubbins of cartilages persisting near the ventral midline (Fig. 1H, arrowheads). Ectopic cartilage nodules are frequently positioned ventrally, such as near the basihyal (Fig 1H, asterisks, Table 1). For the dorsal opercle bone in the second arch, we observe both loss and gain (expansion) phenotypes (Kimmel et al., 2003b), while the corresponding ventral bone, the branchiostegal ray posterior, is frequently missing (Table 1). We also discovered that the plcβ3 mutant phenotype is partially dominant, a characteristic not previously reported (Kimmel et al., 1998; Kimmel et al., 2003a). In plcβ3 heterozygotes there is a variable penetrance of mild facial skeletal defects, similar in nature but much less severe than those of homozygous plcβ3 mutants (Fig. 1 F,I, Table 1, Table 3). These mild facial skeletal defects include joint fusions, reductions in the length of the symplectic region of the hyomandibular cartilage, ventrally-localized ectopic cartilage nodules, and mild changes in the shape of the opercle and branchiostegal ray bones. Both plcβ3 alleles yield skeletal phenotypes with similar penetrance and expressivity.

Fig. 1. plcβ3 mutants have defects in facial skeletal patterning.

(A,B,C) Lateral views of live wild-type, plcβ3 homozygous, and plcβ3 heterozygous larvae at 6 dpf. plcβ3 homozygotes have an open-mouth phenotype and fail to form a swimbladder (white asterisk in A). (D,E,F) Flat mounts of wild-type, plcβ3 homozygous, and plcβ3 heterozygous larvae Alcian-Green stained mandibular and hyoid cartilages at 6 dpf. Homozygotes have fusions at the joint region between the palatoquadrate and Meckel’s cartilage of the mandibular arch and strong reductions in the symplectic and ceratohyal cartilages of the hyoid arch. Meckel’s cartilage is less severely reduced in length and the palatoquadrate is reduced and/or misshaped. Wild-type D/V joint regions are indicated with arrows in (D). Fusions at joint regions in plcβ3 homozygotes are indicated with asterisk in (E). Reductions in symplectic cartilage in plcβ3 heterozygotes are indicated with asterisk in (F). Cartilages are labeled as followed: pq (palatoquadrate), mc (Meckel’s cartilage), hm (hyomandibula), ch (ceratohyal), sy (symplectic), and ih (interhyal). Two bones of hyoid arch are also lightly stained with Alcian Green: op (opercle), bsr_p (branchiostegal ray posterior). The opercle is frequently reduced or expanded in plcβ3 heterozygotes, while branchiostegal ray posterior is frequently absent. (G,H,I) Flatmounts of wild-type, plcβ3 homozygous, and plcβ3 heterozygous larvae doubly stained with Alcian Blue and Alizarin Red for cartilage and bone, respectively. In homozygotes ectopic cartilage nodules are frequently located near basihyal (bh) cartilage (asterisks in H). Ceratohyal cartilages are also +-frequently reduced to nubbins (arrowheads in H). Anterior ceratobranchials (cb) are typically reduced in length. Ceratobranchials are indicated with dots in G. In heterozygotes ectopic cartilage nodules frequently extend from the palatoquadrate cartilage (asterisk in I). Scale bars: 50 μm.

Table 3.

plcβ3 genetically interacts with edn1

Class		Number Scored	Phenotypic Index
			1 (wt)	2 (mild)	3 (moderate)	4 (severe)
plcβ3	edn1
+/+	+/+	31	100%	0%	0%	0%
+/−	+/+	37	78%	22%	0%	0%
+/+	+/−	29	45%	55%	0%	0%
+/−	+/−	41	0%	49%	51%	0%

Offspring of a plcβ3^th210/+ x edn1^tf216b/+ cross were raised to 6 dpf.

The heads of larvae were fixed and stained for cartilage and bone and genomic DNA for genotyping was prepared from the tails. Scoring was done blindly, prior to determining genotype.

“Number Scored” refers to larvae scored. Phenotypic classes were defined as in Table 2.

The skeletal defects we observe in plcβ3 mutants are reminiscent of those in edn1 and furinA mutants, suggesting all three genes may function within a common genetic pathway (Kimmel et al., 1998; Miller et al., 2000; Piotrowski et al., 1996; Walker et al., 2006). In all three mutants, skeletal defects are isolated to the anterior four arches, with cartilages of the posterior three arches having near wild-type appearance. furinA homozygotes are milder in phenotype than plcβ3 homozygotes as furinA homozygotes typically do not have strong reductions of ventral facial cartilages. Similarly, plcβ3 homozygotes are slightly less severe in phenotype than edn1 homozygotes as reductions of ventral cartilages are more penetrant in edn1 homozygotes than in plcβ3 homozygotes.

plcβ3 alleles have lesions in the catalytic domain of Plcβ3

We mapped plcβ3^tg203e to zebrafish chromosome 7 using bulk segregant analysis (Knapik et al., 1996). Finer mapping placed plcβ3^tg203e at 60.7 cM, flanked by the microsatellite markers, Z7958 and Z8540 (Fig. 2A). A zebrafish EST to cct7 also mapped to this interval between Z7958 and Z8540. We confirmed tight linkage of cct7 to she using the primers 5′CAAAGCACAACTTACTTCTACAC3′/5′CCATTCAATCTAAGACTTCTAAG3′ flanking a CA repeat adjacent to cct7. As an insertional zebrafish mutation in cct7 does not have phenotypes similar to those of she mutants, we did not consider cct7 a likely candidate for she, but rather used this positional linkage to look for potential candidates. cct7 is located on Zv4_scaffold645.2 at 51.4 MB on LG7. Nearby at 51.9 Mb, on AL772136 is sequence highly similar to human Plcβ3. We considered plcβ3 an excellent candidate as the ligand Edn1 signals through a receptor (EdnrA) which is coupled to G-proteins of the G_q/G₁₁ family (Ivey et al., 2003). This family of G proteins regulates the function of Plcβ family members, including that of Plcβ3.

Fig. 2. Segregation and sequence analysis of two plcβ3 alleles and plcβ3 MO splice variant.

(A) schmerle (plcβ3) maps to Danio rerio chromosome 7 between microsatellites Z7958 and Z8540. (B) Schematic of protein domains in plcβ3⁺ (wt) and lesions in plcβ3^tg203e and plcβ3^th210. Sequence around lesion sites is aligned with other vertebrate Plcβ3 and with Plc in fly. Dots indicate sequence identity. Abreviations: PH (pleckstrin homology), EF (EF hand), C2 (Protein Kinase C conserved region 2).

(C) Sequence chromatograms of plcβ3^tg203e and plcβ3^th210 around lesion sites in plcβ3. In plcβ3^tg203e a T-to-C mutation results in the transformation of a conserved serine to a phenylalanine in the X domain of the catalytic domain. In plcβ3^th210 a C-to-T mutation results in the transformation of a conserved leucine to proline in the Y domain of the catalytic domain. (D) RT-PCR analysis of plcβ3 mRNA structure in wild-type and splice MO-injected embryos. plcβ3 splice-blocking MO is complementary to exon 4 splice donor site. Primers designed to exon1 and exon5 were used in RT-PCR of uninjected and plcβ3 splice-blocking MO injected embryos at 1 dpf (5′CCGTTGTTACACTGAAGGT3′/5′GCTTTGAGTAAGAAGGTGTTG3′). (E) cDNA sequence comparison reveals that plcβ3 splice-blocking MO variant results from aberrant splicing to a cryptic splice site locate 51 bases 5′ of the normal exon4 splice donor. The plcβ3 MO splice variant results in the deletion of 17 amino acids within the PH domain of plcβ3.

For plcβ3^tg203e a T-to-C mutation results in the transformation of a conserved lysine to proline in the Y domain of the catalytic domain (Fig. 2B,C); segregation with the mutant phenotype was confirmed by PCR genotyping 67 mutant diploid embryos (data not shown). As this transformation results in the loss of a Pvu II restriction enzyme site in mutant embryos, primers 5′GTGAGACATGAACATAGCTG3′/5′GACTTTAACTGTATTGGCTAC3′ were used followed by Pvu II digestion. For plcβ3^th210 a C-to-T mutation results in the transformation of a conserved serine to phenylalanine in the X domain of the catalytic domain (Fig. 2B,C); segregation with the mutant phenotype was confirmed by PCR genotyping 74 mutant diploid embryos (data not shown). Primers 5′CACCTTACCCTCTGAACCTGT3′/5′CCAAGGTAAGATATACACAATG 3′ were used to turn the th210 lesion into a co-dominant polymorphism following Xmn I restriction enzyme digestion.

A splice-blocking morpholino-mediated deletion in the PH domain of plcβ3 partially rescues plcβ3 homozygotes

To demonstrate that a loss of plcβ3 function can cause the mutant phenotypes we observe, we designed both translation- and splice-blocking plcβ3 MOs. We designed a plcβ3 splice-blocking MO to exon 4 donor splice site. Exon 4 encodes part of the N-terminal PH domain, a domain critical for both substrate binding and membrane localization. We tested the effectiveness of the plcβ3 splice-blocking MO in blocking splicing by analyzing the structure of plcβ3 mRNA in uninjected wild-type and plcβ3 splice-blocking MO injected embryos at 1 dpf by RT-PCR. In plcβ3 splice-blocking MO injected embryos, we detect a smaller splice variant of plcβ3 and failed to detect wild-type plcβ3 splice product by RT-PCR (Fig. 2D). Sequencing confirms this splice variant results in the deletion of 17 amino acids within the PH domain of Plcβ3 (Fig. 2E). Our splice MO would therefore be predicted to be effective at interrupting Plcβ3 gene function by affecting substrate binding and membrane localization.

Injections of the plcβ3 splice-blocking MO into wild-type embryos resulted in embryos with a low penetrance of mild facial skeletal defects (phenotypic index 2) (Table 2, Fig. 3B). The mild phenotypes are surprising as our evidence above suggests the splice-blocking MO is effective at blocking Plcβ3 function. These skeletal defects are similar to those we observe in plcβ3 heterozygotes and are less severe than plcβ3 homozygotes (phenotypic index 4). Injections of a translation-blocking MO yielded similar results (Table 2). Co-injections of translation- and splice-blocking MOs into wild-type embryos significantly increased the penetrance of observed phenotypes, but only slightly increased the severity of skeletal defects (Table 2).

Table 2.

Phenotypic Severity Index of plcβ3 mutants and morpholino injected larvae

Class	Number Scored	Phenotypic Index
		1 (wt)	2 (mild)	3 (moderate)	4 (severe)
WT	92	100%	0%	0%	0%
WT + splice MO	86	71%	29%	0%	0%
WT + transl MO	78	78%	22%	0%	0%
WT + splice/transl MO	48	10%	79%	10%	0%
plcβ3−/−	179	0%	1%	2%	97%
plcβ3+/−	148	*	97%	3%	0%
plcβ3−/− + splice MO	20	0%	50%	50%	0%

Morpholinos were injected into the yolk of 1–4 cell stage embryos. At 4–6 dpf larvae were fixed and stained for both cartilage and bone. Scoring was done blindly, prior to determining genotype. ‘Number Scored’ refers to larvae scored. Phenotypic classes are defined as follows:

1 (wt) = phenotypically similar to uninjected wild-type embryos.

2 (mild) = Similar to Fig 1F,I. May have shortened symplectic cartilage, ectopic cartilage, bone fusion, or unilateral joint loss without severe cartilage reduction.

3 (moderate) = More severe than Fig 1F,I, but less severe than Fig 1H. May have bilateral joint losses and mild reduction in ventral cartilage.

4 (severe) = Similar to Fig 1H, with bilateral joint losses and severe ventral cartilage reductions.

^*For the class ‘plcβ3^+/−‘ only larvae with phenotypes are included in table.

Fig. 3. A splice-blocking MO partially rescues plcβ3 homozygotes.

(A–D) Flatmounts of 5–6 dpf wild-type and plcβ3 homozygotes pharyngeal tissues doubly stained with Alcian Blue and Alizarin Red for cartilage and bone, respectively. (A) uninjected wildtype. (B) Injection of plcβ3 splice-blocking MO into wildtype yields only mild phenotypes. Arrow indicates reduction of symplectic cartilage. (C) plcβ3 homozygotes have a loss of jaw joints, misshaped palatoquadrate cartilage, and a strong reduction of symplectic (arrows) and ceratohyal cartilages (arrowheads). (D) Injection of plcβ3 splice-blocking MO into plcβ3 heterozygotes partially rescues cartilage defects. Arrows indicate reduction of symplectic cartilage. Asterisks indicates ectopic cartilage nodule extending from palatoquadrate. Jaw joints and ceratohyal cartilages are strongly rescued. Scale bar: 50 μm.

Since the plcβ3 splice-blocking MO would be predicted to be highly effective at disrupting Plcβ3 function it is important to understand why the plcβ3 MO phenotypes are much less severe than the plcβ3 homozygous phenotypes. The work of Nagano (see Discussion) provides an explanation, namely that the plcβ3 mutations affecting the catalytic domain generates mutant proteins that act as dominant negatives, blocking the function of Plc family members that are also contributors to patterning in wildtype. A prediction of this hypothesis is that if the dominance of the plcβ3 alleles is mediated through substrate binding and/or membrane localization, our splice-blocking MO would relieve this inhibition since it deletes part of the PH domain. We verified this prediction: Injection of the splice-blocking MO into plcβ3 homozygotes (genotyped by PCR) resulted in a strong decrease in severity of facial skeletal phenotypes, including rescue of joint loss and ventral cartilage reductions (Fig. 3C,D, Table 2). Whereas over 90% of uninjected plcβ3 homozygotes have severe phenotypes (phenotypic index 4), none of the splice-blocking MO injected plcβ3 homozygotes had severe phenotypes. This partial rescue strongly supports the hypothesis that the plcβ3 mutations act in a dominant negative manner.

plcβ3 is expressed in the pharyngeal arches and in neuronal cells

The homozygous plcβ3 mutant phenotype is similar to that of Edn1 pathway mutants, suggesting Plcβ3 functions in the Edn1 pathway (Miller et al., 2000; Walker et al., 2006). Since the ligand Edn1 signals through a G protein-coupled receptor (Ednra) localized to neural crest cells ((Clouthier et al., 1998; Nair et al., 2006) data not shown), we predict that Plcβ3 functions just downstream of Ednra and also be expressed within neural crest cells. We compared expression of plcβ3 to known markers of specific cell types by in situ analysis. At 24 hpf plcβ3 mRNA is expressed in a second arch mesenchymal domain similar to dlx2a, suggesting this domain corresponds to postmigratory neural crest cells (Fig. 4A,C). At 24 hpf plcβ3 mRNA expression in the first and posterior arch neural crest domains was very weak and variably detectable (Fig. 4A, data not shown). At 28 hpf plcβ3 mRNA expression in the second arch mesenchymal domain is only weakly expressed, suggesting plcβ3 expression is dynamically regulated. The timing of plcβ3 expression makes sense if Plcβ3 function is required for Edn1-dependent expression of Dlx genes and other transcription factors in neural crest cells (Walker et al., 2006). At 24 hpf and 28 hpf, we also observe plcβ3 expression in neuronal cell types, including a subset of the trigeminal ganglia localized to a dorsal position in the first arch adjacent to the first pouch, similar to a domain of islet1 expression (Fig. 4A,B,C, see arrowheads for comparison). In the trunk, plcβ3 is also expressed in a subset of cells at an appropriate position to be Rohon-Beard spinal sensory neurons, identified by their position relative to the neural tube as compared to islet1 expression (Fig. 4E,F).

Fig. 4. plcβ3 is expressed in the pharyngeal arches and in neuronal cells.

Lateral views of in situ hybridizations in wild-type embryos at 24 (A,C,E,F) and 28 hpf (B,D). Anteior is to the left. At 24 and 28 hpf plcβ3 is expressed in a first arch domain adjacent to the first pouch (A,B), similar to a subset of the islet1 trigeminal ganglion domain (D). At 24 hpf plcβ3 is also expressed in the second arch (A), similar to the dlx2a neural crest domain (C). At 28 hpf plcβ3 is only weakly expressed in the second arch neural crest domain (B). Lateral trunk views at the same axial level of plcβ3 (E) and islet1 (F) RNA in situ hybridizations with wild-type embryos at 24 hpf. plcβ3 is expressed in a subset of cells at appropriate positions to be Rohon-Beard spinal sensory neurons. a1 and a2 label pharyngeal arch 1 and 2. Dotted lines label pharyngeal pouch 1. Arrowheads label cells at position of trigeminal ganglion domain. Scale bar: 50 μm.

plcβ3 genetically interacts with edn1

To directly test whether plcβ3 and edn1 genetically interact, we crossed plcβ3 and edn1 heterozygous fish and assayed offspring for skeletal phenotypes. We observed only mild skeletal phenotypes in offspring with loss of a single copy of either plcβ3 or edn1, with a large percent of the single heterozyotes being phenotypically wild type (Table 3). On the other hand, none of the offspring with a combined loss of one copy of plcβ3 and edn1 were phenotypically wild-type. Rather, in the combined heterozygotes we observed an increase in the penetrance and severity of skeletal defects. These skeletal defects included joint loss, ventral cartilage reductions, ectopic cartilage nodules, and OP/BSR bone phenotypes. These results support our model in which the ligand Edn1 signaling through a G protein-coupled receptor to activate Plcβ3.

plcβ3 is required for edn1-dependent Dlx gene expression

We analyzed expression of other genes in plcβ3 homozygotes to determine if edn1-dependent gene expression was also reduced in plcβ3 homozygotes. We have previously shown that Edn1 signaling is required for the pharyngeal expression of ventrally restricted Distalless (Dlx) transcription factors (Miller et al., 2000; Walker et al., 2006). We assayed for Dlx gene expression at 30 and 36 hpf to determine if plcβ3 is required for both early and late Dlx gene expression. At 30 hpf the expression of dlx5a, dlx6a, and dlx3b were strongly reduced in both the first and second arches in homozygous plcβ3 mutants, while only the ventral expression of dlx2a was reduced (Fig 5. E–H). At 36 hpf the expression of dlx6a and dlx3b remained strongly reduced, but the expression of dlx5a had recovered dorsally but not ventrally in both the first and second arches, and also around the stomodeum (Fig. 5N,O,P). The dlx2a ventral expression domain defect was still evident at 36 hpf (Fig. 5M). In plcβ3 homozygotes we also observed a general persistence of Dlx expression adjacent to pharyngeal pouches. The ventral downregulation of Dlx expression is similar in nature to what we observed in edn1 mutants and suggests that a critical level of early Edn1 signaling that is required for both early and late Dlx expression (Walker et al., 2006) is not reached in homozygous plcβ3 mutants.

Fig. 5. plcβ3 is required for edn1-dependent Dlx gene expression.

Lateral views of in situ hybridizations in wild-type and plcβ3 homozygotes at 30 hpf (AH) and 36 hpf (I–P). (A,E,I,M) dlx2a, (B,F,J,N) dlx5a, (C,G,K,O) dlx6a, (D,H,L,P) dlx3b. At 30 hpf expression of dlx2a is moderately affected in the most ventral domain in plcβ3 homozygotes (E), and at 36 hpf dlx2a expression is clearly reduced in a ventral arch domain in both the first and second arches in plcβ3 homozygotes (M). At 30 hpf the expression of dlx3b, dlx5a, and dlx6a are strongly reduced in both the first and second arches in plcβ3 homozygotes (F,G,H). At 36 hpf expression of dlx6a and dlx3b are still strongly reduced in plcβ3 homozygotes (O,P). At 36 hpf expression of dlx5a has recovered in a dorsal arch domain, but is still strongly reduced in intermediate and ventral arch domains (N). a1 and a2 label pharyngeal arch 1 and 2. Dotted lines label pharyngeal pouches 1 and 2 in (A–P) and bottom of arch 1 and 2 in (F,N,G,O,H,P). Scale bar: 50 μm.

Intermediate and ventral pharyngeal arch domains are miss-specified in plcβ3 mutants

The penetrance of anterior arch joint losses is 100% in both plcβ3 homozygotes and edn1 mutants. In edn1 mutants, expression of bapx1, a gene required for jaw joint formation, fails to be expressed in the intermediate joint-forming domain of the first arch (Miller et al., 2003). Thus, we predicted that bapx1 expression in the intermediate joint forming domain of the first arch also would be strongly down regulated in plcβ3 homozygotes. At 48 hpf the expression of bapx1 within the first arch intermediate domain is strongly reduced in homozygous plcβ3 mutants (Fig. 6G). On the other hand, severe ventral cartilage reductions in arches 1–4 are not as penetrant in plcβ3 homozygotes compared to edn1 mutants (Table 1) (Miller et al., 2000). To explain this difference in penetrance, and also to further support placement of Plcβ3 within the Edn1 pathway, we analyzed the expression of several edn1-dependent transcription factors required for intermediate and ventral arch patterning. We observe an early defect in hand2 expression in plcβ3 homozygotes that recovers significantly by 55 hpf to near wild-type levels in homozygous plcβ3 mutants (Fig. 6H). But, in plcβ3 homozygotes there is a noticable reduction in the amount of ventral tissue that is hand2 positive at 55 hpf. edn1 mutants also show a reduction in ventral arch tissues (Walker et al., 2006). We have previously analyzed the expression of goosecoid (gsc) and barH-like homeobox 1 (barx) in edn1 mutants (Miller et al., 2000; Walker et al., 2006). In homozygous plcβ3 mutants we observe ectopic expression of these genes within the intermediate arch domains, and a reduction in their ventral arch expression (Fig. 6I,J). Finally, we looked at the expression of the skeletal markers runx2b and sox9a in homozygous plcβ3 mutants. Again, we find an expansion of these markers into the intermediate arch domain, and a reduction of their expression ventrally (Fig. 6K,L). Generally, the similarities of expression defects to edn1 mutants and furinA mutants strongly supports plcβ3 being an edn1 pathway gene. As the ventral expression of some of these markers (hand2 and barx1) is not as strongly reduced in plcβ3 homozygotes compared to edn1 mutants, this data fits nicely with the less penetrant severe ventral cartilage reductions we observe in plcβ3 homozygotes.

Fig. 6. Intermediate and ventral pharyngeal arch domains are misspecified in plcβ3 mutants.

Lateral views of in situ hybridizations in wild-type and plcβ3 homozygotes. (A,G) At 48 hpf bapx1 expression in a first arch intermediate domain is strongly reduced in plcβ3 homozygotes. (B,H) At 55 hpf hand2 expression is at near wild-type levels in plcβ3 homozygotes. (C,I) At 48 hpf gsc is ectopically expressed in first and second arch intermediate domains, and strongly reduced in a second arch ventral domain in plcβ3 homozygotes. (D,J) At 48 hpf barx1 expression in the first and second arches is ectopically expressed in intermediate domains and strongly reduced in ventral domains in plcβ3 homozygotes. (E,K) At 55 hpf runx2b is ectopically expressed in a second arch intermediate domain, and strongly reduced in a second arch ventral domain in plcβ3 homozygotes. (F, L) At 48 hpf sox9a expression is strongly reduced in ventral domains of the first and second arches in plcβ3 homozygotes. Arrowheads indicate first arch intermediate domain in wild-type embryos. Arrows indicate second arch intermediate domains in wild-type embryos. Asterisks indicate ectopic expression in intermediate arch domains in plcβ3 homozygotes. a1 and a2 label pharyngeal arch 1 and 2. Scale bar: 50 μm.

Plcβ3 function is required in cranial neural crest cells for intermediate and ventral arch fates

If Plcβ3 functions on the Edn1-pathway, we expect it to function downstream from Edn1, hence in neural crest-derived cells, the targets of Edn1 signaling. To test whether Plcβ3 function is required in these cells, or alternatively in surface ectoderm or endoderm, to pattern the face, we unilaterally transplanted wild-type cells from donor embryos into homozygous plcβ3 mutant host embryos at shield stage (6 hpf) and assayed for rescue of skeletal defects by comparing transplanted side to non-transplanted sides at 5–6 dpf. We limited our analysis to large transplants, in which roughly 25% or more donor tissue contributed to neural crest, ectoderm, or endoderm in host embryo, as assayed at 30–36 hpf. We did not observe rescue of skeletal defects with transplants containing either surface ectoderm (0/12 transplants, Fig. 7B,E) or pharyngeal endoderm (0/14 transplant, Fig. 7C,F). With transplants of presumptive neural crest cells, we observed a strong rescue of skeletal phenotypes on transplanted side of hosts compared to non-transplanted side (12/12 transplants, Fig. 7A,D). The rescued cartilage and bone was fli1:GFP positive, indicating that wild-type fli1:GFP donor tissue contributed to rescued cartilage and bone tissues in plcβ3 mutant host. In addition, we observe rescue of ventral arch tissues in general, including muscle patterning (Fig. 8). These results demonstrate that Plcβ3 function is required in neural crest cells and is not in other cell types to pattern intermediate and ventral arch fates.

Fig. 7. plcβ3 is required in cranial neural crest cells for pharyngeal arch patterning.

(A,B,C) Projections of confocal z-stacks, including fluorescent and bright-field channels, of lateral and orthogonal views of live 36 hpf mosaic host embryos. (A) Cells from the animal pole of Alexa568-labeled sphere staged (6 hpf) wild-type fli:GFP donors were transplanted unilaterally into (A) the neural crest domain, (B) the surface ectoderm domain, (C) the endoderm domain of plcβ3 homozygous hosts. Embryos were screened at 30–36 hpf to determine contribution of donor cells to pharyngeal arches and again at 6 dpf to assess contribution of donor cells to cartilages. Dots indicate bottom of the first pharyngeal pouch. Green florescense indicates presence of fli-positive donor neural crest in transplants in (A) and absence of green florescense indicates absence of neural crest in transplants in (B,C). (D,E,F) Embryos imaged in (A,B,C) were subsequently fixed and doubly stained for both cartilage (Alcian Blue) and bone (Alizarin Red). Non-transplant sides of hosts were used as negative controls for rescue. As brain tissues frequently accompany large neural crest transplants, we also analyzed transplants containing only brain tissue and determined they did not rescue (n=5, data not shown). Scale bars: 50 μm.

Fig. 8. plcβ3 is required in cranial neural crest cells for ventral pharyngeal muscle patterning.

Flat mount in Fig. 7.D was subsequently imaged with DIC to show muscle patterning. On transplant side wild-type neural crest cells partially rescue patterning of several ventral arch muscles. On non-transplant side, muscle fibers are largely disorganized. Muscles are abbreviated as follows: intermandibularis posterior (imp), interhyal (ih), hypohyal (hh) ((Schilling and Kimmel, 1997). Asterisks (*) indicate disorganized muscle fibers on non-transplant side of plcβ3 homozygous host. Scale bar: 50 μm.

plcβ3^−/− neural crest cells sort from wild-type neural crest cells in mosaic embryos

We next performed reciprocal transplants to further test the cell autonomous nature of Plcβ3 requirement in neural crest cells for ventral cartilage formation. We transplanted wild-type and plcβ3^−/− neural crest cells into wild-type fli1:GFP host embryos, and assayed for skeletal phenotypes at 5–6 dpf. We did not observe an induction of mutant phenotypes when plcβ3^−/− neural crest cells were transplanted into wild-type hosts to support our conclusion from the opposite transplant that Plcβ3 function is required within the skeletogenic crest itself. However, we did observe a sorting behavior in which plcβ^−/− donor neural crest cells contributed significantly to dorsal cartilages but not to intermediate and ventral cartilages (Fig. 9B,D,F,H). Moreover, in a mosaic animal, wild-type neural crest cells contributed largely to intermediate and ventral cartilages but not to dorsal cartilages. When wild-type donor neural crest cells were transplanted to a wild-type host, the wild-type donor neural crest cells contributed to cartilages located at any dorso-ventral position within the pharyngeal arch (Fig. 9A,C,E,G).

Fig. 9. plcβ3^−/− neural crest cells sort from wild-type neural crest cells in mosaic embryos.

(A–F) Confocal fluorescent stacks of live 6 dpf mosaic host embryos. (A,C,E) Wild-type Alexa568-labeled neural crest cells were transplanted into wild-type fli:GFP hosts at sphere stage (6 hpf). Wild-type donor cells contributed to (A) hyomandibular (hm), (C) symplectic (sy), (E) interhyal (ih), and ceratohyal (ch) cartilages. (B,D,F) plcβ3 homozygous Alexa568-labeled neural crest cells were transplanted into wild-type fli:GFP hosts at sphere stage (6 hpf). plcβ3 mutant donor cells contributed significantly to (B) hyomandibular, but not to (D) symplectic, (F) interhyal, or ceratohyal cartilages. (G,H) Schematic of cell sorting observed in mosaic transplants. (G) Wild-type to wild-type transplants did not sort, whereas (H) mutant to wild-type transplants sorted. Red indicates donor and green indicates hosts. Scale bar: 50 μm.

The sorting feature of the mosaics suggests that plcβ3^−/− neural crest cells are at a disadvantage in populating the intermediate and ventral arch domain when confronted with wild-type neural crest cells (see Discussion). With very large transplants we do observe a small fraction of plcβ3^−/− neural crest cells contributing to intermediate and ventral cartilages (data not shown).

Discussion

We show that schmerle encodes a zebrafish ortholog of Plcβ3 required for intermediate and ventral pharyngeal arch fates. plcβ3 homozygotes have jaw phenotypes similar to Edn1 mutants, while plcβ3 heterozygotes have similar but much milder facial defects. Edn1 signals through a G protein-coupled receptor to activate Plcβ type enzymes. Our study clarifies which Plcβ family member(s) are critical for facial patterning as none of the single mouse mutants in Plcβ family member(s) yielded skeletal phenotypes. Our analysis of plcβ3 mutant zebrafish demonstrates that plcβ3 is a critical effector of Edn1 signaling in cranial neural crest cells. Furthermore, the dominant negative phenotypes we observe suggest other Plcβ family members can function within the edn1 pathway as well. Perhaps a redundancy for Plcβ function amongst family members extends to mouse facial patterning, explaining why the loss of a single Plcβ family member in mice does not result in facial skeletal defects.

Independent catalytic domain mutations in plcβ3 are partially dominant negative

Two independent plcβ3 alleles, plcβ3^th210 and plcβ3^tg203e, yield similar skeletal phenotypes, and both have lesions in conserved catalytic residues, within the X and Y domains respectively. Heterozygous plcβ3 mutants have mild facial skeletal defects suggestive of a partial loss of Edn1 signaling, while homozygous plcβ3 mutants have severe facial skeletal defects resembling a strong loss of Edn1 function. plcβ3 MO injected embryos have mild phenotypes similar to heterozygous plcβ3 mutants. We show that injection of a splice-blocking MO designed to the N-terminal PH domain in Plcβ3 can partially rescue the facial skeletal defects in homozyogus plcβ3 mutants. How can we understand these results?

To explain, we hypothesize that the partially dominant plcβ3 mutations act in a negative manner due to lesions in conserved residues within the catalytic domains, resulting in catalytically inactive enzymes. Nagano et al. provide evidence to support the dominant negative nature of catalytically inactive PLC enzymes through analysis of a Plcδ4 splice variant, termed ALT III, which lacks catalytic activity due to altered splicing of the catalytic X domain (Nagano et al., 1999). The authors demonstrate that the inactive splice variant acts as a negative regulator not only of Plcδ4, but of other PLC enzymes as well, and that its PH domain is necessary and sufficient for this inhibition. When ALT III was co-expressed with full-length PLCδ4, ALT III inhibited the PLC activity of PLCδ4. Futhermore, the PH domain alone but not a PH domain deletion construct of ALT III could inhibit the PLC activity of PLCδ4. These results show that the PH domain is necessary and sufficient for the inhibitory effect on PLCδ4 activity. ALT III was also found to inhibit the activity of other PLC enzymes, with inhibition most effective against PLCδ type enzymes. As ALT III is expressed at high levels within certain tissues and cell types, ALT III may function as a negative regulator of PLCδ type enzymes within specific developmental and/or cellular contexts. In addition, catalytically inactive forms of PLC-like proteins have also been identified (Plcl1 and Plcl2), suggesting an emerging family of proteins that may function as modulators of important signaling pathways (Otsuki et al., 1999; Takeuchi et al., 2000). Mouse mutants deficient in Plcl1 and Plcl2 have been constructed and initial analysis of these mutants support the hypothesis that catalytically inactive PLC-like genes have important functions as modulators of multiple signaling pathways (Kanematsu et al., 2002; Takenaka et al., 2003; Terunuma et al., 2004).

Nagano’s findings can help explain why we observe dominant phenotypes in plcβ3 catalytic domain mutants. If the dominance of the plcβ3 alleles is mediated through either membrane localization or substrate binding, a deletion in the PH domain could reverse the dominance, as PH domains have clearly been shown to mediate membrane localization (Lemmon and Ferguson, 2000; Razzini et al., 2000; Varnai et al., 2002). Thus, a failure to locate to the membrane could relieve a negative regulatory effect on PLC signaling. In this respect, Plcβ3 membrane localization may be necessary to recruit other proteins to form a signaling complex. If so, a catalytically inactive Plcβ3 with an intact PH domain could possibly form an inactive signaling complex, and thus act in a negative regulatory manner on Plcβ-mediated signaling. We demonstrate that a splice-blocking MO which deletes part of the PH domain of Plcβ3 is effective at rescuing the dominant negative phenotypes of the plcβ3 catalytic domain mutations, strongly supporting our hypothesis that the plcβ3 catalytic domain mutations affect the function of other Plcβ family members.

Previous studies did not identify schmerle (plcβ3) alleles as partially dominant mutations (Kimmel et al., 1998; Kimmel et al., 2003a; Piotrowski et al., 1996). The schmerle mutant has always been recognized as highly variable in phenotype, but the dominance of the mutations was not appreciated until cloning and genotyping revealed that the subtle phenotypes segregated to heterozygotes and the strong phenotypes to homozygotes. In addition, our identification and study of other mutants within the Edn1 pathway has enabled us to recognize the subtle cartilage nubbins and bone fusions in plcβ3 heterozygotes as recurring themes in Edn1 pathway mutants (data not shown) (Miller et al., 2000; Walker et al., 2006).

Plcβ3 functions within the Edn1 signaling pathway

The skeletal phenotype of homozygous plcβ3 mutants is similar to those of furinA and edn1 mutants. We have previously shown that furinA genetically interacts with edn1, and in this paper we show that plcβ3 also genetically interacts with edn1. Furthermore, all three mutants have defects in the same facial skeletal elements but to varying degrees of penetrance and expressivity. Homozygous plcβ3 mutants have a high expressivity of fusions in the joint regions of the first and second arch, and a somewhat milder reduction in the length of the lower jaw (first arch) compared to reduction in the length of the ventral cartilage of the second arch. This phenotype is similar to edn1 mutants but the defects in ventral cartilage reduction are more penetrant in edn1 mutants than in plcβ3 homozygotes (Miller et al., 2000). On the other hand, furinA mutants have a lower expressivity of joint fusions and ventral cartilage reductions in both the first and second arch compared to plcβ3 heterozygotes. Thus, furinA mutants are phenotypically milder than homozygous plcβ3 mutants in terms of facial skeletal defects. Injections of endothelin1 MO at different doses can phenocopy the full range of phenotypes of all of these mutations (Kimmel et al., 2003a; Miller and Kimmel, 2001). We hypothesize that homozygous plcβ3 mutants have a stronger defect in Edn1 signaling than furinA mutants. Analysis of edn1-dependent Dlx gene expression in furinA and homozygous plcβ3 mutants supports this hypothesis. At 30 hpf the arch expression of dlx5a, dlx6a, and dlx3b are strongly reduced in homozygous plcβ3 mutants, as we show here, but only moderately reduced in furinA mutants (Walker et al., 2006). At 36 hpf the expression of these Dlx genes have significantly recovered in furinA mutants, but is still strongly reduced in homozygous plcβ3 mutants. These results demonstrate that a critical level of early Edn1 signaling, that is required for late Dlx gene expression, is reached in furinA mutants but not in homozygous plcβ3 mutants or edn1 mutants.

However, analysis of other edn1-dependent transcription factors, particularly that of hand2, distinguishes between homozygous plcβ3 and edn1 mutants in terms of severity of expression defects. We have previously shown by mutational analyses that hand2 is an edn1-dependent transcription factor required for the formation of the ventral cartilages of the anterior arches (Miller et al., 2000; Miller et al., 2003). hand2 expression at 55 hpf (i.e., at a very late stage of patterning) is at near wild-type levels in homozygous plcβ3 mutants (Fig. 6H), but is strongly reduced in edn1 mutants (Walker et al., 2006). This reduction in edn1 mutants, but not in homozygous plcβ3 mutants, suggests that based on hand2 expression, edn1 mutants have stronger defects in Edn1 signaling than homozygous plcβ3 mutants. It is interesting that homozygous plcβ3 mutants have near wild-type levels of late hand2 expression, because homozygous plcβ3 mutants typically have severe reductions of the ventral cartilage of the second arch. Before our study it was possible to understand the ventral cartilage loss in edn1 mutants as being entirely due to downregulation of hand2 (Miller et al., 2003). However, this lack of correlation between hand2 expression and ventral cartilage phenotype suggests that other factors in addition to hand2 are required for ventral cartilage formation. Part of the reduction in ventral cartilage formation in plcβ3 homozygotes may be due to a general reduction in tissues which express hand2, rather than a reduction in the hand2 signal per se.

In addition, the most ventral cartilage in the second arch persists in plcβ3 homozygotes as cartilage nubbins, suggesting a graded requirement for Edn1 function along the length of the ventral cartilage. This concept of graded requirement for Edn1 function ties into our previous model in which intermediate arch domains are more sensitive to reductions in Edn1 signal than ventral arch domains (Miller et al., 2003; Walker et al., 2006). The study of Edn1 pathway mutants with varying reductions in Edn1 signaling has enabled us to understand more fully the requirements for Edn1 function in specific skeletal elements.

Plcβ3 function is required autonomously in cranial neural crest cells for pharyngeal arch patterning

We show that Plcβ3 functions autonomously in cartilage-making tissue. When wild-type neural crest cells are transplanted into a homozygous plcβ3 mutant host we observe a strong rescue of the facial skeleton on the transplanted side of the plcβ3 mutant host at 6 dpf. If Plcβ3 functions within the Edn1 signaling pathway as we argue above, it is downstream of Edn1, which is expressed by signaling cells, and also downstream to the membrane spanning G-coupled receptor, Ednra, expressed in neural crest cells. The timing of expression of plcβ3 in neural crest cells fits nicely with this hypothesis.

Our rescue experiments also reveal that Plcβ3 functions non-autonomously in neural crest cells to pattern surrounding tissues, possibly including muscle patterning. This function of Plcβ3 in patterning muscles supports our earlier hypothesis that Edn1 signaling specifies ventral neural crest cell fates, including ventral cartilages, bones, and connective tissues, and that these ventrally-specified neural crest cells then signal back to mesoderm to allow correct patterning of ventral muscles (Miller et al., 2000). Similarly, Kontges and Lumsden hypothesize specifically that neural crest-derived connective tissue which inserts into muscle patterns mesoderm-derived muscle fibers (Kontges and Lumsden, 1996). Availability of a cell-autonomous mutation acting in neural crest will let us examine this proposition critically, by constructing mosaics in which only muscle connective tissue is mutant.

Mosaic sorting of plcβ3 mutant neural crest cells

When Clouthier and coworkers (2003) generated chimeric mouse embryos using Ednra mutant embryonic stem cells, they observed that Ednra mutant cells were excluded from the caudal-ventral aspects of the pharyngeal arches. Furthermore, in a few embryos in which Ednra mutant cells did populate this region of the arch, they did not contribute to structures that are derived from this region (Clouthier et al., 2003). Our mosaic experiments with plcβ3 mutants yields similar results in terms of exclusion of plcβ3 mutants cells from ventral arch, but unlike Clouthier et al. we observe that plcβ3 mutant cells can contribute to ventral cartilages. Indeed, in plcβ3 homozygotes we also observe a small percent of mutants with ventral cartilages. These differences could be due to plcβ3 mutations not fully inactivating the Edn1 pathway.

We hypothesize that Edn1 signaling promotes expansion of cells in the ventral arch to support Edn1-dependent arch elongation (Walker et al., 2006). In situ analysis with PCNA and BrdU labeling has shown the ventral arch to be a highly proliferative zone (Plaster et al., 2006) so the expansion may be brought about by cell division, as we are examining in experiments in progress. Plcβ3 function may be required cell-autonomously for neural crest cells to proliferate and contribute significantly to ventral cartilages, in response to Edn1 signaling. Mosaic analysis of plcβ3 mutant neural crest cells in an edn1 mutant host will enable us to further determine if the mosaic sorting we observe when plcβ3 mutant neural crest cells confront wild-type crest cells is Edn1-dependent, as we propose.