This study was carried out in accordance with the recommendations of the animal study guideline of the Animal Experimentation Ethical Committee of Chongqing Medical University. The protocol was approved by the Animal Experimentation Ethical Committee of Chongqing Medical University in Chongqing, China. Embryo in vitro fertilization, culture, whole mount in situ hybridization, mRNA preparation and microinjection were carried out as previously described (Shi et al., 2009, 2014). The probes for whole-mount in situ hybridization including Chordin, Xbra, Wnt8, Fgf8, Msx1,Pax3, Zic1, Zic5, Slug, Six1, MyoD were applied as described (Smith et al., 1991; Sasai et al., 1994; Nakata et al., 2000; Li et al., 2006; Guemar et al., 2007; Zhang et al., 2014). For enteric neuron labeling, intestines were first manually dissected out from stage 40 embryos which were freshly fixed with 4% paraformaldehyde. Then the enteric neurons were stained through in situ hybridization with N-tubulin probe (Oschwald et al., 1991).

The cranial cartilage staining was performed as previously described (Shi et al., 2014). For immunohistochemistry, embryos were collected at stage 40 and fixed with paraformaldehyde. The frozen samples were sectioned in 10 μm thickness, and stained with β-tubulin III (Sigma T8660). Total RNA samples were extracted with Trizol kit (Tiangen) and reversely transcribed with Fermentas RevertAid First Strand cDNA synthesis kit (Thermoscientific 1622). The RT-PCR primers for Fgf8, Wnt8, Dlx3, Dlx5, AP2α, Pax3, Msx1, Zic1, Zic5, Slug, Hes5.2a, Hes5.2b, ESR1, and Histone4 (H4) were listed in Table 1.

T-Leukemia cell lines Jurkat were cultured with 2 mM ketamine for 10 h and collected for WB. The Antibodies used were Notch-1 antibody (Cell Signaling 3608, 1:1000), GAPDH (Enogene E12-052-4, 1:5000). For Notch ubiquitin assay, 5 μM MG132 (Sigma) was added with or without ketamine into the media, Jurkat cells were then lysed and precipitated with Notch-1 antibody (1:50) and then bound with protein A/G (Santa Cruz sc-2003). Anti-ubiquitin (Santa Cruz sc-8017) antibody were applied to detect ubiquitinated proteins.

Different fragments (100 bp–4 kb) upstream of Xenopus tropicalis Zic5 gene were cloned and linked with XhoI and MluI restrictive site into pGL3-basic to generate the firefly reporter constructs. The primers for different lengths of X. tropicalis Zic5 upstream regulatory sequences are: reverse primer XhoI: 5′-ccctcgagtgtctgcctcccaactct; -4042 forward: 5′-cgacgcgttgagatggcgagtaggct; -1950 forward MluI: 5′-cgacgcgttccttattagtgtatata; -532 forward: 5′-cgacgcgttgcacaactataggtctatt; -286 forward: 5′-cgacgcgtctccaaactttctacaagtg; -200 forward: 5′-cgacgcgtcagccagccaatcagaaaag; -200m forward: 5′-cgacgcgtttctaattaaggattgaaaagcgggcctcc; -186 forward: 5′-cgacgcgtgaaaagcgggcctcctgcc. To mutant canonical CSL binding site, mutant kit (Takara R401) as well as template construct pXzic5 -4042/-29 was applied according to the manufacture with primers: for 3950 site forward 5′-atgtattgcttcgaccttggcattttgg, reverse 5′-gtttctacattttccaaagttgtgcaaag; for 65 site forward 5′-ggacagacctacggaaggaataagc, reverse 5′-ctccagcgattggcagaaagcg.

To measure the reporter activity in embryos, 12.5 pg constructed firefly plasmid, 2.5 pg renilla construct and 500 pg NICD mRNA or 500 pg LacZ mRNA (for control) were co-injected into both side of the dorsal blastomeres at the 4-cell stage. The embryos were then harvested at the neural plate stage (stage 15), divided into three groups (>10 embryos each group), lysed and analyzed using Dual-Luciferase Reporter Assay System (Promega).

SHSY-5Y cell line was transfected with pcDNA4-NICD or pcDNA empty vector. Nuclear protein was extracted. EMSA was performed as described previously with minor adjustment (Wang et al., 2011). Briefly, 20 μg protein was incubated with IRDye700-labeled binding oligo (5′-gcctgacagccagccaatca) or mutant oligo (5′-gactaacaaccatccaaaca) respectively, and the gels were analyzed using an Odyssey system (LI-COR Biosciences). Unlabeled wild-type and mutant oligonucleotides at X100 molar excess were applied for the competition assay.

Previous data indicates that 2 mM ketamine applied in culture yields approximately 0.4% accumulation, i.e., 8 μM inside embryo (Cuevas et al., 2013), which is comparable to the anesthetic concentration in human plasma (Dong et al., 2014). 2 mM or similar culture concentration was commonly chosen to evaluate the embryonic ketamine exposure in most studies (Kanungo et al., 2013; Felix et al., 2014; Lantz-McPeak et al., 2015; Robinson et al., 2015). To examine ketamine’s effects on early embryonic development, we exposed Xenopus embryos to 2 mM ketamine-MBS culture media from pre-gastrulation (stage 7) to early neural plate (stage 13.5), and then cultured in MBS media until harvest time. Upon ketamine incubation, pigment cells in trunk diminished and eyes exhibited developmental abnormities (Figure 1A). Ocular pericytes located in retina ganglionic cell layer (GCL) and inner nuclear layer (INL) are derived from the NC (Trost et al., 2013). Further analysis showed disorganization of GCL, INL, and outer nuclear layer (ONL) (Figure 1B DAPI stain, white arrow) and loss of β-tubulin III expression in cell layer boundary (Figure 1B, white arrowhead) and lens (Figure 1B, asterisk) expression, indicating abnormal differentiation of lens and retina. At the tailbud stage, ketamine caused enlargement of intestinal system combining loss of enteric nervous cells (Figure 1C, black arrowhead), a mega-colon like (Hirschsrpung disease) phenotype (Figure 1C) and marked reduction or even complete loss of all the cranial cartilage elements including Meckel’s (Mk), ceratohyal (ch), basihyal (bh), and branchial (ba) cartilages as demonstrated by alcian blue staining (Figure 1D). These data clearly indicate that ketamine impairs NC development, causing neurocristopathies.

Induction is the first event in NC genesis. To investigate the effect of ketamine on NC induction, we examined the expression pattern of the NC specification gene Slug. The criteria for judging the expression pattern are based on two aspects, expression area and intensity. Compared to control, decreased expression area or signal intensity was determined as ‘reduced expression level.’ At the early neural plate stage (stage 14), when the NC has been induced, control embryos express a normal Slug expression pattern (in 32/32 embryos) (Figure 2A). However, ketamine at a 0.36 mM culture concentration showed reduced Slug expression (36/37). The inhibition of Slug expression increases in a dose dependent manner. At a 2 mM ketamine concentration, Slug expression was severely reduced (41/41) (Figure 2A). Morphogens from the mesoderm are critical for inducing the NC (Klymkowsky et al., 2010). Thus, we examined mesoderm development by in situ hybridization assay with mesoderm markers on the ketamine-incubated and control embryos prior to NC induction (stage 11.5) (Figure 2B). There was no significant difference between ketamine and control groups in the expression pattern of four mesoderm maker genes including Chordin (100% normal expression for ketamine, n = 56; and 98.41% normal expression for control, n = 63), Xbra (98.48% normal expression for ketamine, n = 66; and 100% normal expression for control, n = 46), Wnt8 (95.92% normal expression for ketamine n = 49; and 98.08% normal expression for control, n = 52), and Fgf8 (100% normal expression for ketamine, n = 77; and 100% normal expression for control, n = 55) (Figure 2C). These results suggest that ketamine does not affect mesoderm at the beginning of NC induction. These data indicate that ketamine blocks NC induction independent of mesoderm genesis.

Neural crest induction requires high Wnt and intermediate BMP signaling input (Steventon et al., 2009). At the onset of gastrulation, animal cap, a pluripotent cell group located in at the animal pole of Xenopus embryos, can develop into almost any cell type given the proper signaling stimulation in vitro. In Xenopus animal caps, co-injection of Wnt8 and BMP4 truncated receptor (tBR) mimics endogenous NC stimulation and results in the expression of NC genes (Zhang et al., 2014). To determine the targeting molecules mediating ketamine’s effect on NC induction, we first examined gene expression in the wnt8-tBR co-injected animal cap system. The default of animal cap cells is an epidermis fate, characterized by a high BMP signaling niche (Linker et al., 2009; Montagner et al., 2016). BMP target genes Msx1 (Tribulo et al., 2003), Dlx3, Dlx5 (Luo et al., 2001; McLarren et al., 2003) and AP2α (Luo et al., 2002, 2003) are involved in neural plate border determination and are normally expressed in control animal caps. Co-injection of 500 pg Wnt8 mRNA and 500 pg tBR mRNA successfully triggers animal cap cells to express the secreted genes Fgf8, Wnt8, and neural border genes including Zic1, Pax3, Zic5 as well as NC specific genes Slug (Figure 3A). In the presence of ketamine, Wnt8-tBR induced animal caps markedly reduces the expression of Zic5 and Slug (Figure 3A). Zic5 is expressed in the NC area and plays an indispensable role in the induction process (Nakata et al., 2000). Since embryos undergo NC induction in the late gastrula (approximate at stage 12.5), we examined the NC markers at this critical time with a 1–2 mM ketamine exposure or 0.8 nmol direct injection. At stage 12 the pre-induction state, neither control nor ketamine-treated embryos expressed Slug. At stage 13, all of the NC regulatory network genes were expressed in control embryos, whereas Zic5 and Slug transcripts were reduced in embryos cultured in the ketamine-containing media or directly injected with ketamine. However, the Zic5 and Slug expression levels from whole embryos returned to normal at the beginning of NC migration (stage 16). The recovery NC gene expression may be due to the feedback of complicated GRNs (Figure 3B). Consistent with RT-PCR data from animal caps and whole embryos, in situ hybridization results showed that ketamine treatment did not affect most NC induction-related genes including Fgf8 (97.96% normal expression, n = 49), Wnt8 (100% normal expression, n = 77), Msx1 (100% normal expression, n = 66), Pax3 (97.56% normal expression, n = 41), and Zic1 (96.15% normal expression, n = 52) (Figures 3C,F). However, the expression of Zic5 and the induction marker gene Slug were markedly inhibited by ketamine, with 97% (n = 70,) and 100% (n = 43) of embryos having inhibited expression respectively (Figures 3D,F). Finally, exogenous injection of Zic5 mRNA at two-cell stage significantly rescued the ketamine-induced Slug inhibition (63.64% of normal expression, n = 44) (Figures 3E,F). Our data suggest that ketamine disrupts NC induction by inhibiting Zic5 expression.

To investigate the upstream signaling pathways mediating ketamine-induced Zic5 inhibition during NC induction, we screened the most common NC induction morphogens including Wnt, Fgf, and Notch. Previous work showed that prospective paraxial mesoderm or Fgf8 induces Zic5 expression (Monsoro-Burq et al., 2003). Our results show that Fgf8 and its downstream NC inducing gene Msx1 (Monsoro-Burq et al., 2005) were normally expressed and not affected by ketamine treatment (Figure 3). However, ketamine exposure markedly reduced the expression of Notch target genes Hes5.2a, Hes5.2b, and ESR1 (Figure 4A). Somite labeling with MyoD and Six1 transcripts showed slightly reduced signal (Figure 4B) and a shortened embryonic body axis (Figure 1A) that mirrors phenotypes seen in Notch-deficient embryos (Jen et al., 1997). To analyze ketamine’s effect on Notch signaling, the human T-cell leukemia line Jurkat was used. 2 mM Ketamine exposure for 10 h increased the amount of ubiquitinated Notch (Figure 4C) and promoted Notch protein degradation (Figure 4D). To ask whether Zic5 functions as a downstream target gene of Notch signaling, we assayed Zic5’s response to high and low Notch environment during NC induction in Xenopus embryos. Blocking Notch pathway with microinjection of Delta-stu mRNA (Revinski et al., 2010) at the 4-cell stage markedly inhibited Zic5 expression (Figure 4E). Activation of Notch signaling by injection of NICD (Fryer et al., 2002) induced Zic5 ectopic expression in whole neural plate and NC area (Figure 4E).

To investigate the molecular mechanism underlying Notch’s effect on Zic5, we first examined whether the Zic5 gene contains Notch response element (NRE) in its upstream regulatory region. A 4013 bp containing sequence, between -4042 and -29 bp upstream of the translation start codon (designated as +1), for the X. tropicalis Zic5 gene was cloned into pGL3-basic to generate a firefly luciferase gene reporter construct pXzic5-4042/-29. Sanger sequencing confirmed the insertion of -4042 to -29 as identical to 117367902–117371914 of chromosome 2 (Xenbase/X. tropicalis Gbrowse 9.0). This 4 kb sequence contains two putative canonical Notch signaling CSL binding elements (GTGGGAA) in -65 to -58 (chromosome 2, 11731879–11731885) and in -3950 to -3944 (chromosome2, 11731971–11731977) upstream of Zic5. To check the promoter activity and the involvement of these two CSL cis-acting binding elements, pXzic5 (-65M)-4042/-29 carrying the -65 binding site mutation to CCTACGG, pXzic5 (-3950M)-4042/-29 carrying the -3950 binding site mutation to CCTACGG, and pXzic5 (DM) -4042/-29 carrying both binding site mutations to CCTACGG were generated for reporter assays (Figure 5A). NICD mRNA or LacZ mRNA (mock stimuli), and pXZIC5-4042/-29 or relative 4 kb insertion containing CSL mutant firefly constructs were co-injected into both dorsal blastomeres (that target the neural plate and NC area) at 4-cell stage. A Renilla luciferase construct was also co-injected into embryos as the normalization control. Upon NICD stimulation, pXzic5-4042/-29 (1.59 ± 0.23 RLU), pXzic5 (DM)-4042/-29 (1.51 ± 0.06 RLU), pXzic5 (-3950M)-42/-29 (1.49 ± 0.07 RLU) and pXzic5 (-65M)-42/-29 (1.37 ± 0.13 RLU) reveal significant high luciferase expression level (P < 0.05) compared with negative control (LacZ) stimulated pXzic5-4042/-29 (0.05 ± 0.01 RLU) group in Xenopus Embryos. Transcriptional activity exhibited no statistical difference among the 4 kb wild-type and the putative CSL-binding site mutant constructs (Figure 5A). Our data suggest that there is functional NRE within region -4042 to -29, however, the two putative canonical CSL binding sites did not account for Notch-activated Zic5 transcriptional activation.

To identify the NRE, a series of deletion plasmids, covering various regions of this 4 kb fragment were cloned. Figure 5B showed that NICD significantly activated the luciferase report plasmids containing -4042 bp (2.49 ± 0.62 RLU), -1950 bp (2.36 ± 0.45 RLU), -532 bp (2.73 ± 0.063 RLU), -286 bp (2.30 ± 0.62 RLU), and -200 bp (2.48 ± 0.42 RLU) upstream of Zic5 ORF with similar activation. A further deletion of 15 bp lead to a sharply reduced luciferase activity in pXzic5-186/-29 (0.187 ± 0.099RLU) (P < 0.05) (Figure 5B). Therefore, the 15 bp sequence (5′-ACAGCCAGCCAATCA) located between -200 and -186 bp upstream of Zic5 was identified to contain a functional NRE. To confirm the results, we generated a new plasmid pXzic5-200M containing mutations in the NRE (Figure 5C). The Zic5 upstream 200 bp reporter plasmid containing the 15 bp NRE was strongly activated by NICD (0.073 ± 0.0069 RLU), and the mutations in the 15 bp sequence to disrupt the NRE (0.007 ± 0.0009 RLU) abolished the activation of Zic5 by NICD (Figure 5D).

Electrophoretic mobility shift assay (EMSA) was performed to further examine whether Notch signaling activation promotes the binding of the transcriptional factors to the 15 bp NRE. Nuclear proteins with or without NICD-transfected were extracted from SHSY5Y cells. NICD-transfection significantly increased the amount of the protein-NRE DNA binding complex (Figure 5E). The competitor probe containing the same non-fluorescently labeled 15 bp sequence reversed the fluorescent binding signaling to background level, and a mutant probe had little effect to compete with the binding (Figure 5E). These results unequivocally demonstrate that Notch activates Zic5 expression via the newly identified 15 bp NRE.