High glucose environment inhibits cranial neural crest survival by activating excessive autophagy in the chick embryo

High glucose levels induced by maternal diabetes could lead to defects in neural crest development during embryogenesis, but the cellular mechanism is still not understood. In this study, we observed a defect in chick cranial skeleton, especially parietal bone development in the presence of high glucose levels, which is derived from cranial neural crest cells (CNCC). In early chick embryo, we found that inducing high glucose levels could inhibit the development of CNCC, however, cell proliferation was not significantly involved. Nevertheless, apoptotic CNCC increased in the presence of high levels of glucose. In addition, the expression of apoptosis and autophagy relevant genes were elevated by high glucose treatment. Next, the application of beads soaked in either an autophagy stimulator (Tunicamycin) or inhibitor (Hydroxychloroquine) functionally proved that autophagy was involved in regulating the production of CNCC in the presence of high glucose levels. Our observations suggest that the ERK pathway, rather than the mTOR pathway, most likely participates in mediating the autophagy induced by high glucose. Taken together, our observations indicated that exposure to high levels of glucose could inhibit the survival of CNCC by affecting cell apoptosis, which might result from the dysregulation of the autophagic process.


Results
Exposure to high glucose levels lead to developmental defects in the chick craniofacial skeleton. In our previous study, we demonstrated that the incidence of neural tube defects increased with increasing levels of glucose exposure 22 . Exencephaly might occur if the neural tube defects occur in the cranial neural tube. Of course, not only neurogenesis but also cranial osteogenesis is involved in exencephaly. The craniofacial skeleton of the vertebrate head is a complicated system of interconnected bones and is derived primarily from the cranial neural crest cells. We first examined chondrogenesis and osteogenesis in the chick skull using alcian blue/alizarin red s staining ( Fig. 1A-C; N = 10 embryos in each group). In these experiments, we used L-Glucose as osmolarity control for high D-glucose treatment as previously described 24 . The chondrogenesis in the chick skulls in the high glucose-treated group was not different from that in the control group. There appeared to be some developmental defects in the frontal (FR) bones (indicated by the black arrow in Fig. 1B,C) and parietal (PA) bones in the high glucose-treated group (indicated by the white arrow in Fig. 1B,C). Double alcian blue/alizarin red s staining can sometimes obscure the alizarin red s staining in ossified skeletal structures, so the alizarin red s staining was repeated in chick skulls treated with the same high levels of glucose ( Fig. 1D-F). In comparison to the control, an obvious defect in parietal (PA) bone development was found, as indicated by the black arrow in Fig. 1F, which could be distinctly observed at high magnification (Fig. 1F'). The incidence of parietal (PA) bone developmental defects was 60% of the total number of chick embryos treated with high levels of D-glucose ( Fig. 1G; 0% in control group, N = 0/15 embryos; 18% in L-Glc group, N = 3/17 embryos; and 60% in D-Glc group, N = 12/20 embryos, P < 0.05). Additionally, we measured and analyzed the area of parietal bones in both control group and high glucose-treated groups. In presence of glucose, the area of parietal bones decreased obviously ( Fig. 1H; Control = 13230 ± 1018 μ m 2 ; L-Glc = 11780 ± 1618 μ m 2 ; D-Glc = 6556 ± 1924 μ m 2 , P < 0.05).

Exposure to high glucose levels impaired the generation of cranial neural crest cells. Descendants
of the cranial neural crest cells are the primary source of the bones that form the face and the skull 25 . Because we observed a defect in the parietal bones in the presence of high glucose levels, we determined the generation of cranial neural crest cells labeled via HNK1 (migratory neural crest) fluorescent staining following exposure to high glucose levels. The expression of HNK1 in whole-mount chick embryos showed that cranial neural crest cell generation was significantly impaired by D-glucose compared to the control ( Fig. 2A-C). We measured the HNK1 + area per section on serial transverse sections through the heads of multiple treated embryos ( Fig. 2D; Control = 133.2 ± 19.7μ m 2 , N = 8 embryos; L-Glc = 124.7 ± 28.5μ m 2 , N = 10 embryos; D-Glc = 43.1 ± 7.0μ m 2 , N = 10 embryos, P = 0.0015), the tendency of HNK1 + migratory neural crest cells to be reduced was also shown in the sections ( Fig. 2A'-C') at the levels indicated by the white dotted lines in Fig. 2A-C. Another gene expressed in migratory neural crest cells is Slug 26 . Using in situ hybridization to detect Slug in the control and high glucose treated HH10 chick embryos, we found that the production of Slug + migratory cranial neural crest cells was also restricted following the D-glucose treatment (Fig. 2E-G). In the transverse sections of the Slug + in situ stained embryos, we are likewise able to demonstrate a reduction in Slug + migratory cranial neural crest cells in the presence of D-glucose (Fig. 2E'-G'). The emigration areas of the Slug + migratory cranial neural crest cells on sections quantifiably reflected this trend ( Fig. 2H; Control = 93.9 ± 11.6 μ m 2 , N = 7 embryos; L-Glc = 80.1 ± 6.6 μ m 2 , N = 7 embryos; D-Glc = 62.3 ± 5.2 μ m 2 , N = 8 embryos, P = 0.0353). We measured the expression of neural crest markers such as Pax3, Sox9, FoxD3 in the cranial portions of the chick embryos using real-time quantitative PCR following treatment with high levels of glucose. The expression of Pax3 and Sox9 were down-regulated, whereas the expression of FoxD3 was up-regulated ( Fig. 2I-K). We also compared the mRNA expression of neural tube markers such as Pax6, BMP4 and EMT marker like Msx1. We found that in presence of high glucose, the expression of Pax6 was inhibited obviously, while, the elevated glucose concentration had no significant effect on the expression of BMP4, Msx1 (Fig. 2L-N). All of these data suggest that exposure to high glucose levels has a negative impact on development of early neural system, especially on the generation of cranial neural crest cells in chick embryos. crest cells induced by high glucose treatment. We performed immunofluorescent staining for Pax7 to label both the pre-migratory and migratory neural crest cells ( Fig. 3A-C). We measured the Pax7 + cell number per section on serial transverse sections through the heads of multiple treated embryos. Again, we found that the production of Pax7 + cranial neural crest cells was inhibited by D-glucose in comparison to the control ( Fig. 3D; Control = 144 ± 3,   The chick embryos were exposed to simple saline (control) and high glucose (L-Glucose and D-Glucose) for 48 hours before being harvested for Pax7, AP-2α and pHIS3 immunofluorescent staining. L-Glc = 157 ± 12, D-Glc = 114 ± 9, P = 0.0177, N = 7 embryos, respectively). Immunofluorescent staining for pHIS3 was carried out simultaneously ( Fig. 3A'-C'). We analyzed the number of Pax7 + -pHIS3 + neural crest cells and found that D-glucose exposure did not inhibit the proliferation of cranial neural crest cells ( Fig. 3E; Control = 1.8 ± 0.3, L-Glc = 1.8 ± 0.5, D-Glc = 2.0 ± 0.5, P = 0.9214, N = 7 embryos, respectively).

The apoptosis of the cranial neural crest cells is increased in the presence of high glucose levels.
High glucose levels did not seem to have a strong effect on cell proliferation, so we considered whether high glucose levels affected the apoptosis because both proliferation and apoptosis are abundant during embryonic development. Pax7 expression was detected using immunofluorescent staining to confirm the cranial neural crest cells ( (Fig. 4E'-G') to display the neural crest cell morphology. We calculated the ratio between the bright Hoechst cells (apoptotic cell) and the total number of Hoechst positive cells and the ratio between cells that were strongly stained with PI and the total number of Hoechst positive cells to distinguish apoptotic from necrotic cells. The results showed that both apoptosis and necrosis were elevated in the presence of high glucose levels ( Fig. 4H; Apoptotic cells of total: Control = 11.07 ± 1.676%, L-Glc = 18.53 ± 0.9128%, D-Glc = 20.59 ± 1.701%, P = 0.0015; Necrotic cells of total: Control = 6.216 ± 1.040%, L-Glc = 12.66 ± 2.017%, D-Glc = 18.00 ± 1.672%, P = 0.0011; Control: N = 6 explants, L-Glc: N = 7 explants, D-Glc: N = 8 explants). To explore the involved mechanisms in cranial neural crest cell apoptosis, we employed Pifithrin-μ, which can specifically inhibit P53-dependent mitochondrial pathway of apoptosis. With the same strategy, we found that the c-PARP + neural crest cells were obviously decreased in the presence of Pfμ (Supplementary Figure-1A-H, P < 0.001), which indicates that cell apoptosis was induced by D-glucose in a P53-dependent manner. Additionally, using HEK293 cell, we tried to find a common mechanism of cell death induced by glucose. The HEK293 cell line was employed to detect the protein expression of apoptosis-related genes using Western blot. We demonstrated that high glucose treatment increased the expression of P53, pro-Cas7, Cleaved-Cas7, pro-Cas3 and Cleaved-Cas3. These results suggested that elevated glucose levels could trigger cell apoptosis through a P53-dependent manner (Supplementary Figure-1I). We propose that apoptosis is involved in the inhibition of cranial neural crest cells that is induced by exposure to high glucose levels.
Exposure to high glucose levels elevated the expression of autophagy-associated genes in neural crest and HEK293 cells. Cell apoptosis might play an important role in the reduction of the cranial neural crest cells induced by high glucose levels, and dysfunctional autophagy can cause cell apoptosis 27 . Therefore, we wanted to study whether autophagy is involved in the inhibitory effect. Primary cultures of cranial neural crest cells were used because it is easier to detect autophagy-associated gene expression in cell cultures than in embryo samples. We observed that the emigration of neural crest cells from neural tube tissues in the presence of D-glucose was reduced compared to the control (Fig. 5A-C), and the area statistics of neural crest cell emigration showed the same trend ( Fig. 5D; Control = 3.4 ± 0.4 mm 2 , N = 10 explants; L-Glc = 3.2 ± 0.6 mm 2 , N = 10 explants, D-Glc = 2.4 ± 0.1 mm 2 , N = 12 explants; P = 0.0312), confirming the inhibitory effect of high glucose levels on the generation of neural crest cells. In addition, bright/round cells were increased in the D-glucose treated group, indicating that there are more dead cells in the presence of high glucose levels. AP-2α + cells were decreased when the primary cultures of neural crest cells were exposed to high glucose levels ( Fig. 5E-H, AP-2α + cells versus total cells: Control = 81.05 ± 3.988%, L-Glc = 66.56 ± 3.124%, D-Glc = 40.57 ± 3.762%; N = 6 explants in each group, P < 0.001). Our observations in primary cultures of neural crest cells cultured in the presence of high glucose levels are similar to the in vivo embryo experiments. We determined the LC3B expression because LC3B is an important gene in activating autophagy and found elevated LC3B expression and autophagosome-like particles around the cell nuclei in the presence of D-glucose compared to the control ( Fig. 5I-K, I'-K'). Furthermore, Western blots also showed that the high glucose treatment enhanced LC3B protein expression in the chick embryos (Fig. 5L). In addition, we exposed HEK293 cells to high levels of glucose and performed Western blots to detect autophagy-related gene expression. We found increased expression of Beclin-1 and Atg5-Atg12, slightly decreased expression of FL-Atg5, and elevated expression of truncated Atg5. Both LC3B I and LC3B II expression rose (Fig. 5M). We employed transmission electron microscopy (TEM) to observe the autophagosomes in HH10 chick cranial crest cells to verify that autophagy was induced by elevated glucose levels. Compared to the control, we found more autophagosomes in the D-glucose treated cells (Fig. 5N-O). Atg5 cleavage mediated by Calpain is a well-known cell type-independent switch between autophagy and apoptosis 20 . In summary, autophagy is activated in the   presence of high glucose treatment, implying that it might play role on the reduction of cranial neural crest cells partly through truncated Atg5.

Functional evidence for the involvement of autophagy in the reduction of neural crest cells induced by high glucose levels.
We demonstrated that there were changes in autophagy-associated gene expression when embryos or cells exposed to high glucose. We then investigated whether modulating autophagy in the presence of high glucose could directly affect the production of cranial neural crest cells using either Hydroxychloroquine (CLQ, autophagy inhibitor) or tunicamycin (TM, autophagy agonist) 28 as schematically shown in Fig. 6A,B. Either CLQ or TM soaked beads were implanted in the head folds, and simple saline soaked beads were implanted on the opposite side as a control (Fig. 6A,B). The HNK1 immunofluorescent staining was performed to label neural crest cells as the embryos developed until HH10. The result showed that more cranial neural crest cells were produced when autophagy was inhibited with CLQ beads in comparison to the control beads ( Fig. 6C; HNK1 area of control side: Con = 1.1 ± 0.05, N = 5 embryos; CLQ = 2.5 ± 0.3, N = 6 embryos, P = 0.0021). This could be clearly observed in the transverse sections (Fig. 6C'-C"; HNK1). On the contrary, fewer neural crest cells were produced when autophagy was activated with TM beads (Fig. 6D; HNK1 area of control side: Con = 1.1 ± 0.05, N = 5 embryos; TM = 0.6 ± 0.02, N = 6 embryos, P < 0.001). This could also be observed in the transverse sections (Fig. 6D'-D"). This verified that autophagy activation could inhibit neural crest production while autophagy inhibition could increase neural crest production in the presence of elevated glucose (Fig. 6E).
High glucose-induced autophagy is dependent on the ERK rather than the mTOR pathway. The PI3K/Akt/mTOR pathway is known to play a very important role in the autophagy process 29 . The activation of the MEK/ERK pathway also has a critical impact on cytoprotective autophagy 30 . Therefore, we wanted to determine whether PI3K/Akt and ERK signaling are involved in the regulatory mechanism. We determined the levels of protein expression of phosphorylated (p-) Akt and Akt, p-ERK, and ERK in chick head cells in response to high glucose treatment by performing Western blots. Glucose distinctly up-regulated p-Akt and p-ERK expression (Fig. 7A). The relative expression of p-Akt and p-ERK to β -actin was measured ( Fig. 7B-C, P < 0.001), suggesting that the ERK signaling could be activated due to high glucose exposure. The activation of Akt indicated that PI3K/Akt signaling was activated, and which could be an upstream regulator of mTOR. We exposed primary cultures of cranial neural crest cells to high glucose levels and either Rapamycin, an inhibitor of mTOR, or 3-MA, a type III PI3K inhibitor, for 48 hours in vitro to investigate the direct role of mTOR in elevated glucose-induced autophagy 28 . We found that the emigration of neural crest cells from neural tube tissues in the presence of either D-glucose or D-glucose with Rapamycin were similar, indicating the inhibition of mTOR had no effect on neural crest abundance. Nevertheless, the emigration of neural crest cells exposed to D-glucose and 3-MA was obviously reduced compared to the D-glucose group (Fig. 7D upper panel), and the area statistics of the neural crest cell emigration reflected this trend ( Fig. 7E; D-Glc = 2.4 ± 0.1 mm 2 , N = 12 explants; D-Glc + RAPA = 2.5 ± 0.5 mm 2 , N = 6 explants, P = 0.7836; D-Glc + 3-MA = 1.5 ± 0.2 mm 2 , N = 9 explants, P = 0.0002). In addition, we observed that D-glucose and 3-MA significantly suppressed the abundance of neural crest cells that detached from the cultured cranial neural tubes in vitro (Fig. 7D lower panel), suggesting that the inhibition of PI3K signaling impaired cell survival during treatment with high levels of glucose. Taken together, the results suggest that ERK signaling rather than the mTOR pathway is the major signaling pathway involved in elevated glucose-induced autophagy.

Discussions
High glucose environment is associated with a high risk of fetal malformation. However, very little is known about the influence of hyperglycemia on cranial neural crest cell production. In this study, we found defects in partial bone development following treatment with high levels of glucose, especially treatment with D-glucose (Fig. 1G). L-glucose exposure also causes parietal bone defects in a few cases. This effect might be due to high glucose-induced osmotic pressure, which also occurred in our high salt-exposure experiments 31 . However, under normal circumstances, osmolarity is stable in the human body. The osteogenesis of the skull is completed through intramembranous ossification. The cranial neural crest cells are the principal components that contribute to the mesenchymal cells, where ossification will occur 25 . Therefore, we investigated the early stage of cranial neural crest cell production in the experiments described here.
HNK1 is expressed in migratory neural crest cells and was employed to label the migratory cranial neural crest cells. We found that D-glucose significantly inhibits cranial neural crest cell abundance ( Fig. 2A-D). And the production of Slug + cranial neural crest cells demonstrated the same trend ( Fig. 2E-H) as was observed using HNK1. We also measured the expression of Pax3, Sox9, FoxD3, Pax6, BMP4 and Msx1, which are also associated with the generation of the neural crest: Pax3, Pax6 and Sox9 expression were decreased, but FoxD3 expression was increased following exposure to high glucose levels, while, the elevated glucose concentration had no significant effect on the expression of BMP4, Msx1 (Fig. 2I-N). This might suggest that high glucose treatment leads to a reduction in cranial neural crest cells by influencing the expression of Slug, Pax3 and Sox9. Our current observations are similar to the findings of Wentzel that Pax3 mRNA levels were inhibited by high glucose exposure in both cranial and trunk neural crest explant cultures 13 .
The next question we addressed was how the cranial neural crest cell production was restricted by exposure to high glucose levels. Neural crest cells possess some stem cell characteristics; they continue proliferating and differentiating while migrating to their final destination. Thus, we addressed whether the proliferation of cranial neural crest cells was affected by exposure to high levels of glucose. Because Pax7 could label the migratory neural crest cells and the dorsal side of the neural tube, we enumerated the proliferating cranial neural crest cells (Pax7 + -pHIS3 + cells) (Fig. 3A-E). We found that high glucose levels caused a reduction in the number of detached cranial neural crest cells; similar results are obtained using the migratory cranial neural crest marker, AP-2α (Fig. 3F-I). However, it is rather remarkable that the number of both AP-2α + and pHIS3 + neural crest cells were similar to the controls (Fig. 3J), suggesting that migratory cranial neural crest cell proliferation is not affected by high glucose treatment. We next examined the possibility that the presence of high glucose levels induces apoptosis in cranial neural crest cells.
Again, Pax7 was employed to label the cranial neural crest cells (Fig. 4A-C), and cleaved-PARP (c-PARP, cell apotosis marker) was used to detect apoptotic cells (Fig. 4A'-C'). The number of c-PARP + -Pax7 + cells increased in the presence of high levels of glucose. This result demonstrated that cranial neural crest cells undergo more apoptosis following treatment with high levels of glucose (Fig. 4D). Cell death consists of apoptosis and necrosis. Propidium iodide strongly stains dead cells and we analyzed propidium iodide and Hoechst33258 double positive cells in primary cultures of neural crest cells to determine whether necrosis was also elevated in the presence of high glucose levels ( Fig. 4E-G,H). Coincidentally, Cederberg et al. (2003) reported that high-glucose exposure-induced cell death of neural crest derived cells was responsible for the developmental disorder of neural crest-derived structures in diabetic rats embryos 8 . Utilizing specific P53 inhibitor, Pifithrin-μ , we found that elevated glucose concentration could induce apoptosis in a P53-dependent manner. At the meantime, we compared the Pax7 + cell number between D-glucose and D-glucose combined with Pfμ treated groups. The Pax7 + cells decreased significantly in the D-glucose combined with Pfμ treated group (Supplementary Figure-1G, P < 0.05). P53 plays a critical and complex role in regulating apoptosis, necrosis and autophagy, which can promote either cell death or cell survival under stress. Activated P53 in nuclear can induce autophagy. When apoptosis is compromised, autophagy can promote cell death 32,33 . To further understand the mechanisms of high glucose induced cell death, more effort is required.
We next addressed the cause of the elevated levels of cell death among the cranial neural crest cells that were exposed to high levels of glucose. In our previous studies, we showed that high glucose treatment could cause excessive ROS generation 22 and that the excess ROS might in turn damage the cells through dysfunctional autophagy 34 . Hence, we measured the expression of the autophagy-associated gene, LC3B, in primary cultures of cranial neural crest cells after verifying the failure of AP-2α + neural crest cells to detach from cultured neural tubes in the presence of high levels of glucose (Fig. 5A-H). Following treatment with high levels of glucose, we detected elevated LC3B expression via both immunofluorescent staining and Western blot assays (Fig. 5I-L). To further confirm the correlation between autophagy and high glucose treatment, we exposed HEK293 cells to high glucose and used western blot to show that the expression of Beclin-1 and LC3B II were elevated and that Atg5-Atg12 complexes were increased, implying autophagy activation. Truncated-Atg5 expression was increased while FL-Atg5 expression was decreased slightly. Calpain mediated Atg5 cleavage is a cell type-independent switch between autophagy and apoptosis 20 . The expression of autophagy-associated genes suggests the activation of autophagy in the presence of high levels of glucose, and the excess autophagy induced by high glucose provoked cell apoptosis via truncated Scientific RepoRts | 5:18321 | DOI: 10.1038/srep18321 Atg5. The experiments in which CLQ or TM beads were implanted were designed to verify the functional role of autophagy in the cranial neural crest (Fig. 6A,B). As might be expected, we found that the inhibition of autophagy with CLQ could enhance the production of cranial neural crest cells (Fig. 6C-C", E), whereas the activation of autophagy with TM lessened the production of cranial neural crest cells (Fig. 7D-D", E), indicating that there is a functional correlation between autophagy and high glucose exposure. These results are in accordance with the report by Adastra, KL et al. 16 , in which they identified a variety of pathways for activating autophagy in mouse embryos and oocytes in response to a hyperglycemic environment.
Altering either the external or the cellular environment could activate autophagy through different signaling pathways. The MEK/ERK and PI3K/Akt/mTOR signaling pathways play vital roles in regulating cellular autophagy 29,35 , and Class III PI3K plays an essential role in the initiation of autophagosome formation 36 . When we measured the phosphorylation of Akt and ERK in HH10 chick heads, the up-regulation of p-Akt and p-ERK indicated the activation of PI3K/Akt and ERK signaling by exposure to high glucose levels ( Fig. 7A-C). We combined the specific inhibitor of mTOR, Rapamycin, with D-glucose in primary cultures of cranial neural crest cells and found that the production of cranial neural crest cells was not significantly affected compared to the D-glucose group. The survival of neural crest cells was significantly inhibited by D-glucose with 3-MA (Fig. 7D,E). These results suggest that the activation of autophagy induced by high glucose is independent of mTOR and that PI3K/Akt signaling possibly plays a key role in cell survival in the presence of high glucose levels. In summary, our current experimental data suggest that exposure to high glucose levels leads to excessive autophagy in the cranial neural crest cells through an mTOR independent pathway. ERK signaling could be a major pathway involved in this. Furthermore, the excess autophagy could lead to cell apoptosis in cranial neural crest cells, ultimately reducing the production of cranial neural crest cells in the presence of high levels of glucose (Fig. 8). Further experiments are certainly required to determine the regulatory mechanism at the molecular level. There is no doubt that the more we understand about the pathological and molecular mechanisms underlying cranial neural crest cell dysplasia in response to high glucose levels, the better we will be able to prevent and treat NCC diseases (neurocristopathies).

Methods
Chick embryos and Glucose application. Fertilized leghorn eggs were acquired from the Avian Farm of the South China Agriculture University (Guangzhou, China). Early developing chick embryos are not considered to be animals, so serious ethical issues are not evoked 37 . The fertilized eggs were incubated in a humidified incubator (Yiheng Instruments, Shanghai, China) at 38 °C and 70% humidity until the desired Hamburger-Hamilton (HH) stage of chick embryo development was reached 38 . The glucose treatment in ovo was performed according to the methods described by Hinds group 24 . Before experimentation, the air chamber of the eggs was marked under an egg candler. The shell above the air chamber was carefully removed, and 50 μ l of 50 mM D-Glucose (dissolved in 0.72% sodium chloride, Sigma, G7528), L-Glucose (osmolarity control, Sigma, G5500) or chick simple saline (0.72% sodium chloride) was injected into the air chamber of the chick embryos every day from 0-day until 12-day. The experiments were performed in triplicate with 10 eggs in each group, and the surviving embryos were harvested for skeleton staining.
For early chick embryos, the same methods for treatment in early chick (EC) 39 cultures as those we described previously were employed in this study 18 . Briefly, HH1 embryos were exposed to either D-glucose or L-glucose containing medium (90 mg glucose/10 ml medium) and then incubated at 37 °C with humidity until HH10. To understand the role of P53 in apoptosis induced by high glucose, EC cultured embryos were treated with D-glucose combined with Pifithrin-μ (20μ M, SIGMA, #P0122) until HH10. Embryos exposed to D-Glucose only were set as control. Figure 8. A proposed model that depicts the potential mechanisms by which high glucose exposure leads to defects in cranial neural crest cell generation. High levels of glucose might trigger autophagic disturbances and promote ERK activation. The expression of the neural crest development-related genes Pax3, Sox9 and Slug were also adversely influenced. It is well known that elevated levels of glucose can also induce autophagy through ER stress, although we did not address this mechanism in the current study. In summary, we propose that excessive cell autophagy induced by high glucose levels is the key factor that induces increased apoptosis in the cranial neural crest. Alcian blue/alizarin red staining of whole embryos. The 12-day (E12) chick embryos mentioned above were stained with alcian blue and alizarin red dyes to visualize the craniofacial skeleton, as previously described 40 . Briefly, the embryos were fixed in 95% ethanol for 2 hours and then the skin and viscera were carefully removed and post-fixed for 1 day. Next, the embryos were stained in 0.1% alcian blue and/or alizarin red (Solarbio, Beijing, China) dyes in 70% ethanol for 1 day and then cleared in 25% glycerol/0.5% KOH for 3 days. Finally, the embryos were treated in a graded series of glycerol solutions. Then, the cranial skeleton was photographed using a stereomicroscope (Olympus, MVX10).
In situ hybridization. The whole-mount in situ hybridization of the chick embryos was performed according to the standard in situ hybridization protocol 42 . Digoxigenin-labeled probes were synthesized against Slug. The whole-mount stained embryos were photographed, and then, frozen sections were prepared from them by sectioning at thickness of 16 μ m.
Transmission electron microscopy. The control and glucose treated HH10 chick embryos were fixed with 2.5% glutaral in 0.1 M PBS for 2 h, and then, the chick heads were dissected. The samples were sent to the TEM Laboratory of Sun Yat-sen University. The embedding, ultrathin sectioning and staining were performed by professional technicians and examined using a Tecnai G 2 Spirit Twin (FEI, USA). The observation of autophagosomes using TEM was according to the guidelines for monitoring autophagy 28 .
RNA isolation and Quantitative PCR. The total RNA was isolated from the HH10 chick heads (N > 20 embryos in each group) using a Trizol kit (Invitrogen, #15596018) according to the manufacturer's instructions.
Primary Culture of NCCs and cell staining. The NCCs were prepared from the neural tube (at head level) of the chick embryos, according to the methods previously described 45 . Briefly, the fertilized chick eggs were incubated until the 7-9 somite stage (HH 9-9+ ). The neural tubes were dissected from the head of the embryos and explanted into 3.5 mm dishes in DMEM (100 μ l/dish, containing 5.5 mM D-glucose, Life, #11885-092) for 6 hours at 37 °C and 5% CO 2 to allow the explants to adhere. A few of the NCCs migrated from the neural tubes after the incubation and then 1 ml of medium containing 50 mM D-glucose/L-glucose or with 200 nM Rapamycin (dissolved in DMSO, < 0.1%, final concentration) or 5 mM 3-Methyladenine (3-MA, dissolved in simple saline) was introduced into the cultures for 48 hours. The migration area and was measured using Image-Pro Plus 7.0 software as we described previously 46 .
To determine apoptosis and cell death, the Hoechst/Propidium Iodide (PI) double staining experiments were carried out according to the methods described previously 47 . Briefly, after incubation, the culture medium was removed, and the neural crest cells were rinsed with pre-warmed PBS and stained in 1 ml of PBS containing 10 μ g/ ml Hoechst33258 (Life, H1398) and 10 μ g/ml PI (Life, P1304MP) at 37 °C for 15 min. After staining, the cultures were washed thoroughly with PBS and fixed with 4% paraformaldehyde at room temperature for 15 min. Following a wash, F-actin (1:1000, CST, #8878) staining was performed at 4 °C overnight. Then, images were taken using an Olympus IX51epi-fluorescent microscope (340 nm, 488 nm and 620 nm respectively). We randomly selected six visual fields (at 400X) from each explant and manually counted the PI + Hoechst + cells or bright blue Hoechst + cells versus total Hoechst + cells using Image-Pro Plus 7.0 software. Bead experiments. The bead experiments were performed according to the methods described by Yamada group previously 43 . Briefly, heparin beads were soaked in Hydroxychloroquine (CLQ, 50 μ M, autophagy inhibitor, Sigma, H0915) or tunicamycin (TM, 4 μ g/ml, autophagy agonist, Millipore, #654380) 28 for 2 h and then implanted into the cranial region of HH8 chick embryos in EC culture, by the side of the head neural tube. The beads soaked in simple saline were used as a control (as shown in Fig. 6A,C). The embryos were continually incubated until