Rnf220/Zc4h2-mediated monoubiquitylation of Phox2 is required for noradrenergic neuron development

Noradrenergic (NA) neurons are located in both the central nervous system (CNS) and the peripheral nervous system. In the CNS, these neurons mainly reside in the brainstem and can be classified into six groups by their locations, including: the locus coeruleus (LC) group (A6), the subcoeruleus group and four other groups (A1, A2, A5 and A7) (Robertson et al., 2013). LC-NA neurons (NA neurons located in the LC) represent the vast majority of noradrenaline-releasing neurons in the CNS, with widespread projection in the brain. Dysfunction of the NA system is implicated in both psychiatric and neurodegenerative disorders, such as cognition decline, anxiety, depression, and Parkinson's and Alzheimer's diseases (Sara, 2009; Itoi and Sugimoto, 2010; Szot, 2012).

LC-NA neurons are generated from dorsal rhombomere 1 (r1) around embryonic day (E) 10.5 and migrate tangentially and ventrally to the dorsolateral pons at the late embryonic stages (Robertson et al., 2013). The genetic cascade controlling the development of LC-NA neurons has been extensively documented. FGF8 expressed in the mid-hindbrain border (isthmus), and BMPs and Wnts expressed in the roof plate, are essential for the specification of NA progenitors (Vogel-Hopker and Rohrer, 2002; Holm et al., 2006; Tilleman et al., 2010). Notch/Rbpj signaling in the ventricular zone is also involved, shown by findings that LC-NA neurons significantly increase in number when this signaling pathway is blocked (Shi et al., 2012). On the other hand, deletion of Ascl1 (Mash1) leads to a failure of NA neuron generation (Hirsch et al., 1998). In addition, the roles of transcription factors expressed in postmitotic LC-NA neurons have also been revealed (Flames and Hobert, 2011). Among these factors, the homeodomain transcription factors Phox2a and Phox2b have attracted the most attention, because they are selectively expressed in postmitotic LC-NA neurons and directly control the expression of tyrosine hydroxylase (Th) and dopamine-β-hydroxylase (Dbh), two key enzymes that are responsible for noradrenaline biosynthesis. In the LC, Phox2a is activated first and is required for the expression of Phox2b (Morin et al., 1997; Pattyn et al., 1997). The expression of Phox2b is transient and downregulated at E13.5 in the LC (Pattyn et al., 2000), whereas ongoing expression of Phox2a is required for maintenance of NA differentiation. LC-NA neurons cannot be generated in mice with constitutive knockout (KO) of either Phox2a or Phox2b (Pattyn et al., 1997; Flames and Hobert, 2011). Deletion of Phox2a and Phox2b simultaneously at the adult stage also impairs the maintenance of LC-NA neurons (Coppola et al., 2010).

ZC4H2, an X-chromosome-linked intellectual disability (XLID) gene, encodes a C4H2 type zinc-finger nuclear factor, the mutation of which causes arthrogryposis multiplex congenita, intellectual disability and other defects (Hirata et al., 2013; May et al., 2015; Zanzottera et al., 2017). We previously reported that Zc4h2-mediated control of BMP/Smads signaling is required for Xenopus early neural development (Ma et al., 2017). Rnf220 is a RING finger domain ubiquitin E3 ligase that is responsible for the polyubiquitylation and degradation of Sin3B, a global regulator of gene transcription, which serves as an essential scaffold protein of the Sin3/HDAC corepressor complex (Kong et al., 2010), and involved in the regulation of canonical Wnt signaling through modulating USP7-mediated deubiquitylation of β-catenin (Ma et al., 2014). Recently, we reported that Zc4h2 and Rnf220 show similar expression patterns in the spinal tube and are required for ventral spinal tube patterning through modulation of Shh/Gli signaling in the mouse (Ma et al., 2019a; Ma et al., 2019b). Our and Kim's studies have shown that Zc4h2 interacts with Rnf220, and enhances the stability of Rnf220 both in vitro and in vivo (Kim et al., 2018; Ma et al., 2019a). However, the functions of Zc4h2 and Rnf220 in the development of the brain remain unknown.

In a microarray profiling analysis, we saw that the expression of the LC marker Dbh was greatly reduced in the Rnf220 KO mouse brain (data to be published elsewhere), suggesting possible deficiency of LC-NA neurons in the absence of Rnf220. Here, we analyze in detail the roles of Zc4h2 and Rnf220 in the development of LC-NA neurons in the mouse. Zc4h2 and Rnf220 are both expressed in developing LC-NA neurons, and KO of either gene leads to a complete loss of expression of LC-NA neuron-associated genes at late embryonic stages, although the initiation of these genes are normal at the beginning of development. Furthermore, we provide evidence that the Zc4h2 and Rnf220 complex interacts with and monoubiquitylates Phox2a and Phox2b, promoting transactivation of Phox2a/Phox2b and expression of Th and Dbh, the hallmark of normal NA neuron differentiation.

The expression of Zc4h2 was examined in combination with molecular markers of NA neurons in mouse embryos at different embryonic stages. At E10.5, Dbh transcripts were observed in the dorsal r1, where LC-NA neurons are born (Hirsch et al., 1998), and these Dbh⁺ cells also expressed Phox2a and Phox2b (Fig. 1A). No Zc4h2 immunoreactivity was detected at this stage (Fig. 1A). At E11.5, Zc4h2 expression was observed in the presumptive LC based on similar locations of Zc4h2⁺ and Dbh⁺ cells, shown by immunostaining of adjacent sections (Fig. S1A,B). In addition, BrdU labeling showed that Zc4h2 was expressed outside the BrdU⁺ ventricular zone (Fig. S1C). At E12.5, Zc4h2 and Phox2a/Phox2b were expressed in Dbh⁺ NA neurons (Fig. 1B), and some Zc4h2⁺ cells without Dbh expression were located ventrolateral to NA neurons (Fig. 1B). At E16.5, Phox2a was expressed only in Th⁺ neurons in the LC (Fig. 1C), and cells with only faint, if any, Phox2b immunoreactivity were located within the LC; in contrast, many cells with strong Phox2b immunoreactivity were located outside (dorsal to) the LC (Fig. 1C). At this stage, intense Zc4h2 signals were observed in the LC and Barrington's nucleus located ventromedial to the LC (Fig. 1C). Collectively, the findings indicated that Zc4h2 is expressed in developing LC-NA neurons.

The role of Zc4h2 in the development of LC-NA neurons were examined in Zc4h2 KO embryos, in which the loss of Zc4h2 was confirmed (Fig. 2A). At E10.5, the expression of Dbh, Phox2a and Phox2b appeared not to be changed in Zc4h2 KO embryos compared with controls (Fig. 2C). At E12.5, however, Dbh expression was significantly reduced, and Phox2a expression was slightly reduced without obvious alterations of Phox2b expression in Zc4h2 KO embryos relative to controls (Fig. 2D). The statistical data demonstrating the decrease of Dbh⁺ neurons in Zc4h2 KO mice is shown in Fig. 2B. At E16.5, neither Dbh nor Phox2a/Phox2b expression was observed in Zc4h2 KO embryos (Fig. 2E). Because the Zc4h2 gene is located on the X chromosome, the female Zc4h2 heterozygote is a chimera with both wild-type cells and Zc4h2 mutant cells because of random X inactivation, as reported previously (Ma et al., 2019a). Indeed, the number of Zc4h2⁺ cells appeared to be reduced in the presumptive rostral pons containing LC-NA neurons in female heterozygotes relative to female controls at E12.5 (Fig. S2A-C). Importantly, Dbh expression was also reduced in the heterozygotes at E12.5 (Fig. S2D). Thus, our data from both mutant and female heterozygotes indicate that Zc4h2 is required for the normal development of LC-NA neurons.

Next, we examined the expression of Rnf220 in developing LC-NA neurons. Rnf220 mRNA was detected in the dorsal r1 at E10.5, and its distribution overlapped with that of Dbh transcripts at E10.5 and E12.5 (Fig. 3A,B). At E16.5, Rnf220 mRNA was strongly expressed in both the LC and Barrington's nucleus, and this pattern was almost identical to that of Zc4h2 in the dorsolateral pons (Fig. 3C). Double immunostaining of Zc4h2 protein and Rnf220 mRNA confirmed the co-localization of the two genes in LC-NA neurons at E11.5 (Fig. S1D).

Successful deletion of the Rnf220 gene was confirmed by in situ hybridization with probes specifically targeting exon 2 (Rnf220 E2) at E10.5 (Fig. 3D). The expression of Dbh, Phox2a and Phox2b was similar in Rnf220 KO and control embryos at E10.5 (Fig. 3E). At E12.5, however, the expression of Dbh and Phox2a was reduced, although the expression of Phox2b was not obviously changed in Rnf220 KO embryos compared with controls (Fig. 3F). The statistical data demonstrating the decrease of Dbh⁺ neurons in Rnf220 KO mice is shown in Fig. 3G. At E16.5, the expression of Dbh, Phox2a and Phox2b was hardly detected in the LC region in Rnf220 KO embryos (Fig. 3H). The expression of other genes associated with LC-NA neurons, including aromatic amino acid decarboxylase (Aadc; also known as Ddc), monoamine oxidase A (MaoA), norepinephrine transporter (Net; Slc6a2) and Th, was also examined, and their expression was completely lost in both Rnf220 and Zc4h2 KO embryos at this stage (Fig. 4A-C). As mentioned above, Zc4h2 and Rnf220 were also expressed in Barrington's nucleus. The expression of corticotropin-releasing factor (Crf; Crh), a marker for Barrington's nucleus (Flames and Hobert, 2011), did not significantly differ in either Rnf220 KO or Zc4h2 KO embryos relative to controls (Fig. 4D). Meanwhile, the expression of peripherin (Peri; Prph), which labels mesencephalic trigeminal nucleus (Me5) neurons located dorsolateral to LC neurons (Barclay et al., 2007), was not changed in Rnf220 KO mice (Fig. 4C). In addition, no expression of Rnf220 was found in the sympathetic ganglion, in which peripheral NA neurons are located, and peripheral NA neurons shown by Th immunostaining were not affected in Rnf220 or Zc4h2 KO embryos (Fig. S3A,B).

Taken together, these results indicate that defective development of LC-NA neurons occurs in Rnf220 and Zc4h2 KO embryos, suggesting that they may closely coordinate to regulate NA neuron development.

Using co-immunoprecipitation (co-IP) assays, we tested whether Rnf220 or Zc4h2 interacts with factors involved in LC-NA neuron development, including Phox2a, Phox2b, Th and Dbh. When expressed alone, Zc4h2, but not Rnf220, pulled down Phox2a and Phox2b (Fig. S4A). When Rnf220 and Zc4h2 were co-expressed, Rnf220 could also immunoprecipitate Phox2a and Phox2b in HEK293 cells (Fig. 5A), suggesting the formation of a ternary complex. Interestingly, an additional band was observed above the normal Phox2a and Phox2b band when Rnf220 and Zc4h2 were co-expressed (Fig. 5A,B, Fig. S4A,B and Fig. S5B), suggesting covalent modification of the proteins.

As Rnf220 is a ubiquitin E3 ligase, we then tested whether Rnf220/Zc4h2 ubiquitylates Phox2a and Phox2b. First, when the E3 ligase-dead W539R mutant was co-expressed with Zc4h2, the additional band of Phox2a or Phox2b was no longer observed (Fig. 5C and Fig. S4C), suggesting that the modification is dependent upon E3 ligase activity. In addition, we also carried out in vivo ubiquitylation assays using HA-tagged ubiquitin in HEK293 cells and found that the modification induced by wild-type Rnf220/Zc4h2 complex was monoubiquitylation (Fig. 5D and Fig. S4D). To identify the lysine residue ubiquitylated by Rnf220, we next mutated all the lysine residues one by one in both Phox2a and Phox2b, and examined the protein band patterns in the presence of Zc4h2 and Rnf220 (Fig. 5E and Fig. S4E). The results showed that K178 in Phox2a and K185 in Phox2b are the lysines that are monoubiquitylated by the Rnf220/Zc4h2 complex, because the mutations of these lysines (Phox2a K178R and Phox2b K185R) totally abolished the specific modified protein bands (Fig. 5F and Fig. S4F).

We also tested whether endogenous Phox2a/Phox2b is monoubiquitylated by Rnf220/Zc4h2 in the catecholaminergic mouse cell line CATH.a. Phox2a and Phox2b antibodies detected two main bands in CATH.a cell lysates. When Rnf220 and Zc4h2 were both knocked down by small interfering (si)RNAs, the intensities of the upper bands decreased (Fig. 5G), suggesting that Rnf220/Zc4h2 monoubiquitylates Phox2a/Phox2b in vivo. The knockdown efficiency of the siRNAs for Zc4h2 and Rnf220 was verified by real-time RT-PCR and western blot analysis (Fig. S5K,L and data not shown). Finally, we examined endogenous Phox2a and Phox2b for potential monoubiquitylation in E18.5 mid-hindbrain tissues from Rnf220 and Zc4h2 KO embryos. Indeed, specific upper bands for Phox2a and Phox2b were observed in hindbrain lysates from control embryos, but the intensities of these bands were dramatically reduced in both Rnf220 KO and Zc4h2 KO embryos (Fig. 5H,I). Together, the above data strongly suggest that Phox2a and Phox2b can be monoubiquitylated by the Zc4h2/Rnf220 complex. Interestingly, for Phox2a, the modified upper band appeared to be a doublet in both CATH.a cells and brain lysates, suggesting the presence of an additional modification or isoform of Phox2a in vivo.

To explore the functional consequences of Phox2a and Phox2b monoubiquitylation by the Rnf220/Zc4h2 complex, we first examined the DNA-binding ability of Phox2a/2b through electrophoretic mobility shift assay (EMSA) gel shift assays. We found that Phox2a and Phox2b proteins from HEK293 cells with co-expression of the wild-type Rnf220/Zc4h2 complex showed enhanced DNA-binding ability compared with those from HEK293 cells with co-expression of ligase dead mutant Rnf220/Zc4h2 complex or control plasmids (Fig. 6A and Fig. S5A-C). Note that wild-type Phox2a and Phox2b and their KR mutants showed similar DNA-binding abilities (Fig. S5A,B), and Rnf220/Zc4h2 co-expression did not affect the DNA-binding ability of the two KR mutants (Fig. S5D,E). To examine the transcriptional activity of Phox2a and Phox2b, we conducted a human DBH promoter-driven reporter assay. It showed that the transactivity towards the expression of luciferase reporter was enhanced by wild-type Phox2a and Phox2b but not their KR mutants (Fig. 6B and Fig. S5F). We also examined the transcriptional activity of Phox2a and Phox2b by measuring Th and DBH expression levels in CATH.a cells. Co-expression of Zc4h2 and Rnf220 enhanced Phox2a and Phox2b transactivity towards Th and DBH expression, whereas Rnf220 alone or co-expression of Zc4h2 and the Rnf220 ligase mutant did not do so in CATH.a cells (Fig. 6C,D and Fig. S5G,H). The Phox2a K178R and Phox2b K185R mutants, which cannot be ubiquitylated by Rnf220/Zc4h2, exhibited reduced ability to activate Th and Dbh expression, and no enhancement was observed when Rnf220 and Zc4h2 were co-expressed (Fig. 6E,F and Fig. S5I,J). Furthermore, the expression levels of Th and Dbh were decreased when Zc4h2 and Rnf220 were both knocked down in CATH.a cells, but were not significantly changed in the case of single knockdown of Zc4h2 or Rnf220 (Fig. 6G,H), suggesting the possibility that the residual Zc4h2/Rnf220 complex still functions in the monoubiquitylation of Phox2a/Phox2b. Finally, we used in ovo electroporation assays in chick embryos to examine the effect of Rnf220-mediated monoubiquitylation on Phox2 proteins. We first confirmed the expression of overexpressed genes by in situ hybridization (Phox2a, Phox2b and Rnf220) and immunohistochemistry (Zc4h2) in the chick neural tube (Fig. S6). Co-expression of wild-type Phox2a/2b with the Rnf220/Zc4h2 complex induced strongly ectopic expression of Dbh in the chick neural tube, whereas ectopic Dbh expression induced by co-expression of the Phox2a/2b KR mutant with the Rnf220/Zc4h2 complex was less and weaker (Fig. 6I), supporting an in vivo role of Rnf220/Zc4h2 in the regulation of Phox2 transactivity. In addition, the ectopic expression of Dbh was weakly induced in the chick neural tube with overexpression of wild-type Phox2a/2b or their KR mutants but hardly detected in the tube with overexpression of the Rnf220/Zc4h2 complex alone (Fig. S5M). Taken together, these data suggest that Zc4h2/Rnf220-mediated monoubiquitylation of Phox2a and Phox2b favors transactivation of Th and Dbh expression by Phox2a and Phox2b (Fig. 6J).

Following up on our previous studies, we report that Rnf220 and Zc4h2 are required for the development of LC-NA neurons in the mouse brain. Through a mechanistic study, we demonstrated that the Rnf220/Zc4h2 complex monoubiquitylates Phox2a/Phox2b, which is required for the full transcriptional activity of Phox2a/Phox2b. In the absence of either Rnf220 or Zc4h2, the levels of monoubiquitylated Phox2a and Phox2b are reduced, which impairs the differentiation process of LC-NA neurons during embryonic development.

Both Zc4h2 and Rnf220 have similar expression profiles in differentiating LC-NA neurons. Zc4h2 directly binds to Phox2a/Phox2b and is necessary for the monoubiquitylation of Phox2a/Phox2b induced by Rnf220. The initiation of Phox2a, Phox2b and Rnf220 expression in presumptive LC-NA neurons occurs at E10.5, before the initiation of Zc4h2 expression. Thus, the post-translational modification of Phox2a/Phox2b by the Zc4h2/Rnf220 complex does not occur, and the differentiation of LC-NA neurons is unaffected in Zc4h2 and Rnf220 KO embryos at E10.5. As development progresses, Zc4h2 expression is initiated in differentiating LC-NA neurons at E11.5, and modification of Phox2a/Phox2b by the Rnf220/Zc4h2 complex may begin in order to enable proper differentiation of LC-NA neurons. As Phox2a and Phox2b directly regulate the expression of Dbh and Th (Fig. 6 and Fig. S5) and auto-regulate themselves (Rychlik et al., 2005), deficient modification of Phox2a/Phox2b protein by the Zc4h2/Rnf220 complex may account for the downregulated expression of Phox2a and Dbh at E12.5 and the total loss of the expression of LC-NA neuron-associated genes in Zc4h2 and Rnf220 KO embryos at E16.5. However, the terminal fate of LC-NA neurons is still unknown because of the lack of a specific marker to label the mutant LC-NA neurons in Zc4h2 and Rnf220 KO embryos. In addition to Phox2a and Phox2b, the transcription factors Tlx3 (Rnx) and AP-2β (Tfap2b) are also involved in the differentiation or survival of LC-NA neurons (Qian et al., 2001; Hong et al., 2008; Flames and Hobert, 2011). It is unknown whether these factors are also regulated by the Rnf220/Zc4h2 complex. Interestingly, peripheral NA neurons were not affected in Rnf220 or Zc4h2 KO mice. Central and peripheral NA neurons share similar regulatory signals but differ in other pathways (Brunet and Pattyn, 2002; Howard, 2005; Flames and Hobert, 2011). For example, Gata3 is expressed in sympathetic NA neurons and is required for their differentiation and maintenance in embryos and adult mice (Lim et al., 2000; Tsarovina et al., 2010). However, no expression of Gata3 is found in LC-NA neurons (Zhao et al., 2008), and Rnf220 is absent in the sympathetic ganglia. Such differences might account for the different requirements of Phox2a/Phox2b modification by Rnf220/Zc4h2 for the differentiation of NA neurons in the sympathetic ganglia and LC.

Rnf220 has been reported to function through diverse mechanisms in different scenarios; for example, it promotes proteasomal degradation of Sin3B via canonical polyubiquitylation, stabilizes β-catenin through USP7-mediated deubiquitylation in cultured mammalian cells, and changes Glis subcellular location through K63-linked polyubiquitylation during ventral spinal tube patterning (Kong et al., 2010; Ma et al., 2014; Ma et al., 2019b). Zc4h2 is a close partner and stabilizer of Rnf220 (Kim et al., 2018; Ma et al., 2019a). Monoubiquitylation of proteins can alter their activity, structure or localization (Hicke, 2001; Haglund et al., 2003). Although the underlying mechanism remains to be investigated, the present results emphasize a pivotal role of Zc4h2/Rnf220-mediated Phox2 monoubiquitylation in ensuring full transactivation of Th and Dbh expression, whereby Rnf220/Zc4h2 regulates the development of LC-NA neurons. It is possible that the monoubiquitylation of Phox2 recruits other interacting transfactors or favors the DNA-binding ability of the transfactors to facilitate their transcriptional activity; however, this possibility remains to be further investigated.

ZC4H2 is a recently discovered X-linked gene that has been implicated in the regulation of the hippocampal synaptic dendrites, and its mutation leads to brain abnormalities, including intellectual disability (Hirata et al., 2013). As the LC-NA system is a well-known modulator in cognition and other high brain functions (Robertson et al., 2013), it is likely that the impaired central NA system caused by the mutation of ZC4H2 contributes to the development of intellectual disability.

All mice were maintained and handled according to guidelines approved by the Animal Committee of Fudan University and the Animal Care and Use Committee of the Kunming Institute of Zoology, Chinese Academy of Sciences. Rnf220, Zc4h2 KO embryos and Zc4h2 heterozygotes were generated as described in our previous study, and all genotypes were confirmed by PCR (Ma et al., 2019a; Ma et al., 2019b). As similar phenotypes of LC-NA neurons were found in female and male Zc4h2 KO mice, both female and male embryos were used for analysis of phenotypes in Zc4h2 KO mice. In the analysis of Zc4h2 heterozygotes and their controls, only female embryos were used for related analysis.

Immunohistochemical staining and in situ hybridization were performed as described previously (Ma et al., 2019a; Ma et al., 2019b). Briefly, embryos were perfused with 4% paraformaldehyde and postfixed for 24 h. After cryoprotection with 20% sucrose, brains were cut into 12 μm-thick sections using a cryostat (CM1950; Leica) (Coppola et al., 2010). The following primary antibodies were used: rabbit anti-Zc4h2 (1:200; Sigma-Aldrich), rabbit anti-Phox2a (1:300; a gift from Dr A. Pattyn, French Institute of Health and Medical Research, France), rabbit anti-Phox2b (1:500; a gift from Dr A. Pattyn), mouse anti-Th (1:20,000; T1299, Sigma-Aldrich) and rabbit anti-peripherin (1:1000; ab4666, Abcam) antibodies. For double immunostaining of Th and Phox2a, Phox2b and peripherin, sections were incubated with primary antibodies at 4°C overnight, then with a mixture of biotinylated horse anti-rabbit (1:500; BA1100, Vector Laboratories) and Alexa Fluor 488-conjugated donkey anti-mouse (1:500; A21202, Invitrogen) antibodies for 3 h, and finally with Cy3-conjugated streptavidin (1:1000; 016160084, Jackson ImmunoResearch) and Hoechst 33258 (1:2000; 94403, Sigma-Aldrich) for 1 h.

For the BrdU assay, pregnant mice received one pulse of BrdU (100 mg/kg body weight; Sigma-Aldrich) by intraperitoneal injection and were euthanized 1 h later. Brain sections were sequentially subjected to citrate buffer (0.01 M, pH 6.0) at 95°C for 10 min, HCl (2 M) at 37°C for 30 min and sodium borate (0.1 M) at room temperature for 10 min. Treated sections were immunostained with rat anti-BrdU antibody (1:300; OBT0030A, Accurate Chemical & Scientific Corporation) as described above.

RNA probes for detecting Dbh, Phox2a, Phox2b, Net, Crf, Aadc and MaoA were generated based on Allen Brain Atlas (http://portal.brain-map.org). For Rnf220, probes against exon 2 (Rnf220 E2) were generated by RT-PCR.

For double labeling of in situ hybridization and immunostaining, sections were treated with in situ hybridization procedures first. After visualization for mRNA, sections were immunostained with the required antibodies. Images were captured using an epifluorescence microscope (Eclipse 80i, Nikon).

For quantification of Dbh⁺ neurons, Dbh⁺ neurons in four to six sections containing presumptive LC were counted in control and KO embryos by a trained observer who did not know the genotypes (three for each).

Rnf220 and Zc4h2 constructs were described previously (Ma et al., 2019a,b). Mouse Phox2a and Phox2b open reading frame complementary DNAs were obtained by RT-PCR from E10.5 mouse embryo total RNA and then subcloned into the pCS2⁺-N-myc expression vectors. Phox2a and Phox2b KR mutant constructs were created via a site-directed mutation with PCR-driven overlap extension, as described previously (Lim et al., 2000). HEK293 cells were cultured in DMEM containing 10% fetal bovine serum (FBS; Millipore), 100 U/ml penicillin and 100 U/ml streptomycin (Life Technologies). Mouse CATH.a cells were cultured in RPMI-1640 medium containing 8% horse serum (HS, Gibco) and 4% FBS. Both HEK293 and Cath.a cells were transfected using Lipofectamine 2000 (Invitrogen).

For knockdown of Rnf220 and Zc4h2 in CATH.a cells, the following siRNAs were used: siRnf220-1# (siG000055182A, RiboBio), siRnf220-2# (siG000055182B, RiboBio), siZc4h2-1#: 5′-GGTGACCTTTCTTCCACAA-3′ and siZc4h2-2#: 5′-GGAGGGATAGACCTCTGTT-3′.

Trizol reagent (TIANGEN) was used for total RNA extraction and reverse transcription was carried out using a First-Strand cDNA synthesis kit (Thermo Fisher Scientific). The following primers were used for real-time PCR: mouse Rnf220 forward, 5′-TGTGGGCAGAAGCGGATAC-3′, and reverse, 5′-TGTCATCTCCATCCACATCCAG-3′; mouse Zc4h2 forward, 5′-AGCAGGACACAAGGCAGACA-3′, and reverse, 5′-TTGCAAAGAGGGCATATAGG-3′; mouse Th forward, 5′-GGGCATCCTCGATGAGACT-3′, and reverse, 5′-AGAAGAGCCGTCTCAGAGCA-3′; mouse Dbh forward, 5′-CAGCTCTGTCTGGGGGTAGT-3′, and reverse, 5′-GGGGGACGTACTCATCACTT-3′.

For immunoprecipitation, HEK293 cells were lysed in IP buffer [50 mM Tris-HCL (pH 7.4), 150 mM NaCl, 5 mM EDTA (pH 8.0), and 1% Triton X-100] that contained a protease inhibitor mixture (Roche Applied Science) for 30 min on ice, and the lysates were clarified by centrifugation for 15 min at 14,000 g. The protein concentration of each cell lysate sample was determined by bicinchoninic acid (BCA) assay. Immunoprecipitation was carried out with anti-FLAG M2 beads (Sigma-Aldrich). CATH.a cells and E18.5 mouse embryo brain tissues were lysed in SDS lysis buffer. Isolated proteins were subjected to SDS-PAGE followed by western blot analyses. In vivo ubiquitylation assays were conducted as described previously (Ma et al., 2019a). The primary antibodies used for immunoblot were: anti-myc (1:5000; C3956, Sigma-Aldrich), anti-FLAG (1:5000; F7425, Sigma-Aldrich), anti-Zc4h2 (1:1000; HPA027577, Sigma-Aldrich), anti-Phox2a (1:1000, a gift from Dr A. Pattyn), anti-Phox2b (1:1000, a gift from Dr A. Pattyn) anti-Rnf220 (1:1000; HPA049584, Sigma-Aldrich) and anti-α-tubulin (1:5000; 11224-1-AP, ProteinTech).

Sense and antisense oligonucleotides corresponding to the sequences of Phox2-binding domains, containing all the three Phox2 binding domains (PBD1, PBD2 and PBD3) of the human DBH promoter (Seo et al., 2002) were synthesized and labeled by biotin at two ends (Sangon Biotech) with the following nucleotide sequences: forward, 5′-GTGTCATTAGTCCAATTAGAG-3′ and reverse, 5′-CCTCTAATTGGACTAATGACA-3′. The forward and reverse oligonucleotides were annealed, gel-purified and used as probes in EMSA. EMSA and antibody co-incubation experiments were performed according to the instructions of the EMSA kit (Thermo Fisher Scientific).

To construct the human DBH reporter, we inserted the Phox2-binding domain including all the three Phox2 binding sites (PBD1, PBD2 and PBD3) (Seo et al., 2002) into the pGL3-Promoter vector (Promega) within MluI and XhoI sites. We verified all recombinant clones by sequencing. For luciferase reporter assays, HEK293 cells were transfected with the indicated reporter plasmids together with the same TK-Renilla internal control reporter vectors by using the lipofectamine 2000 transfection reagent (Invitrogen) and given fresh medium 6 h after the transfection. Luciferase activity was measured 36 h after the transfection using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions. All assays were performed in at least three independent experiments with a minimum of three replicates.

In ovo electroporation was performed as described previously (Zhu et al., 2007). Fertilized chicken eggs were incubated at 38°C under humid conditions for 50 h to HH stages 12-14 and pCAG-Phox2a/Phox2b/Rnf220/Zc4h2/GFP or pCAG-Phox2a-KR/Phox2b-KR/Rnf220/Zc4h2/GFP (1 µg/µl/plasmid) were injected into the neural tube. After injection, plasmids were electroporated into the neural tube using the Electro Square Porator (ECM830; 12 V, 5 ms, 5 pulses). After 24 h, embryos were dissected out and analyzed.

All of our immunofluorescence and in situ hybridization assays were carried out on at least two indicated embryos. All of our western blot, EMSA, luciferase reporter and real-time RT-PCR assays were conducted at least three times. For Dbh⁺ cell count, independent-sample t-test was performed. For statistical analysis of in vitro experiments, one-way ANOVA evaluations with posthoc Student–Newman–Keuls test were performed. P values of <0.05 were considered statistically significant, and P values of <0.001 were considered statistically very significant. Graphpad Prism 8 software was used for all statistical calculations.

We thank Dr A. Pattyn for providing the Phox2a/Phox2b antibody.

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.185199.reviewer-comments.pdf