Zinc deficiency causes neural tube defects through attenuation of p53 ubiquitylation

Environmental factors influence neural tube closure (Wilde et al., 2014), and failure of closure leads to neural tube defects (NTDs), which are the second-most common birth defect worldwide. NTDs can occur in the cranial region (anencephaly, encephalocele), spinal region (spina bifida) or in combination (craniorachischisis or complex phenotypes) (Copp et al., 2013; Wallingford et al., 2013). Evidence from both experimental and epidemiological studies shows that lower folic acid concentrations are associated with human NTDs and that folic acid or inositol supplementation can reduce the risk of NTDs (Force et al., 2017; Greene et al., 2016; Jin, 2017; Viswanathan et al., 2017; Wilde et al., 2014). Neural tube closure is dependent on the methionine cycle and folate cycle (Leung et al., 2017). Other nutritional factors, such as trace elements, also influence the incidence of NTDs but their function relative to neural tube closure is less known (Copp et al., 2015).

Zinc is an essential trace element from microorganisms to human. Approximately 2800 human proteins can potentially bind zinc in vivo through interactive regions such as zinc-finger motifs, including transcription factors and RING finger domain proteins that play a key role in the ubiquitylation pathway (Andreini et al., 2006). Wild-type female mice and rats fed a zinc-deficient diet yield embryos with NTDs: 4.4% of mouse embryos show exencephaly (Dufner-Beattie et al., 2006) and 3% of rats show spina bifida (Hurley, 1969). Zinc is ingested in food and absorbed into the body via several intestinal zinc transporters. Once absorbed, zinc enters the bloodstream, is taken up into cells and is further distributed to intracellular organelles. Zinc homeostasis is tightly controlled by zinc transporters and metallothioneins. Zinc transporters are divided into two major families: SLC39s/ZIPs, which take zinc into the cells; and SLC30s/ZnTs, which efflux zinc out of cells (Hojyo and Fukada, 2016). Homozygous Zip1^−/− (Slc39a1^−/−) and Zip1^−/−/Zip3^−/− (Slc39a3^−/−) mice fed a zinc-adequate diet (50 ppm zinc/day) show normal reproduction, and embryonic and neonatal growth rates. However, when dams were fed a severely zinc-deficient diet (<1 ppm zinc) beginning on embryonic day (E)8, 16% embryos with parental Zip1 knockout and 54% embryos with parental Zip1/Zip3 double knockout exhibit exencephaly at E14 (Dufner-Beattie et al., 2006). Transport of maternal zinc into the embryonic environment is mediated in part by ZNT1 and knockout of the Znt1 (Slc30a1) gene causes early embryonic lethality (Andrews et al., 2004). These data highlight the association of zinc deficiency with embryonic development and NTDs, but the mechanism by which zinc deficiency affects neural tube closure is unclear.

Epidemiological reports in about 35 papers describe the status of zinc concentration in some human NTD cohorts (Fig. S1). The epidemiology data highlight that lower zinc levels in maternal or fetal tissues are associated with NTDs in Turkey (Çavdar et al., 1991, 1988; Cengiz et al., 2004; Demir et al., 2017; Zeyrek et al., 2009), Iran (Golalipour et al., 2009), China (Yan et al., 2017), Netherlands (Groenen et al., 2003), England (Buamah et al., 1984; Hinks et al., 1989; Leger et al., 1980), Bangladesh (Dey et al., 2010), Mexico (Carrillo-Ponce Mde et al., 2004) and India (Srinivas et al., 2001) (blue frames in Fig. S1). On the other hand, high levels of zinc accumulation are also associated with NTDs (red frames in Fig. S1), suggesting that the balance in zinc levels is important. Moreover, it has been suggested that higher peripheral zinc levels in some NTD cases may be a consequence of a deficiency of zinc uptake, ultimately resulting in zinc deficiency in the disease location. An experimental study has demonstrated that cultured skin fibroblasts from NTD cases with anencephaly and spina bifida show characteristics of decreased zinc uptake (Zimmerman and Rowe, 1983). Importantly, zinc-only or zinc-containing multiple micronutrient supplementation can decrease the risk of NTDs in humans (Czeizel, 1993; Czeizel and Dudas, 1992; Czeizel et al., 1994; Velie et al., 1999). Together, it appears that zinc deficiency has a causative role in some human NTD cases.

This evidence provided the impetus for the experimental studies reported here to explore how zinc deficiency impacts neural tube closure. We have used mouse embryos and primary neuroepithelial cells cultured under zinc-deficient conditions to assess the effect of zinc deficiency on neural tube closure. Time-lapse live imaging methods and molecular studies demonstrate excess apoptosis in embryos during the process of neural tube closure. Moreover, we linked zinc deficiency with the attenuation of p53 ubiquitylation through the zinc-binding protein Mdm2, leading to p53 stabilization and abnormal apoptosis.

To determine the effect of zinc deficiency on neural tube closure, we took advantage of a method to culture mouse embryos ex utero, which allows dynamic visualization of the process of neural tube closure (Massarwa and Niswander, 2013) and a means with which to induce intracellular zinc deficiency using TPEN, a membrane-permeable zinc ion chelator (Cho et al., 2007). The mouse embryo represents an excellent model for human neurulation, as shown by a similar closure process, similar NTD phenotypes and the findings that mouse genes discovered for their role in neural tube closure are correlated with deleterious variants in human NTD cases (Cai and Shi, 2014; Copp et al., 2013; Greene et al., 2009; Li et al., 2018; Wallingford et al., 2013). We used somite-matched E8.5 wild-type mouse embryos, which are just beginning to undergo neural tube closure and which carry a transgene that labels the cellular membranes with myristolyated Venus fluorescent protein (Ven^Myr) for time-lapse live imaging. Using 9-somite embryos at the start of the experiment, 100% of embryos cultured in control media closed the neural tube, and closure occurred by 9 h 18 min±1 h 37 min (n=5; Fig. 1A, Movie 1). In contrast, 100% of embryos at 9 somites treated with 15 µM TPEN (at the lower level of previous studies that used a range of 1-100 µM for cells) (Gao et al., 2017; Rayman et al., 2018; Strand et al., 2013) failed to close the neural tube; there was also a reduction in size of the cranial region and yolk sac, and there appeared to be an abundance of dead tissue in the midbrain-hindbrain region (Fig. 1C, Movie 2, Fig. S2). All of these TPEN-induced defects could be mitigated by the addition of 25 µM ZnSO₄, as 100% of embryos treated with TPEN and ZnSO₄ closed their neural tube (Fig. 1D, Movie 3), although neural tube closure was delayed to 14 h 48 min±3 h 33 min (n=6). At a lower dose of TPEN (10 µM), the neural tube closed (Fig. 1B) but cellular phenotypes were observed, as discussed below. This culture method provides a platform for our subsequent studies to determine the cellular and molecular basis for zinc deficiency-induced NTDs.

To explore further the effect of zinc deficiency on the process of neural tube closure, we quantified various morphogenetic parameters using still images from time-lapse movies. Cranial neural tube closure requires inflection and zippering of the neural folds (Massarwa and Niswander, 2013); therefore, we first examined the potential impact of zinc deficiency on these two morphological processes. Inflection was assessed by measurement of three parameters: perpendicular distance between neural folds, angle of the neural folds and distance between the approaching neural folds (Fig. 1E-G). Zippering was assessed using two parameters: rostral-caudal distance of closing neural folds and caudal zipper point angle (Galea et al., 2017) (Fig. 1H,I). Given that the neural folds in 15 µM TPEN-treated embryos become obscure at around 7 h, we terminated the data collection of this treatment group at 7 h. In control, 10 µM TPEN- and 15 µM TPEN+ZnSO₄-treated embryos, inflection and zippering data points steadily decrease over time (reflecting closure), although the time course is delayed for the TPEN+ZnSO₄ embryos. In contrast, inflection measurements in 15 µM TPEN treatment showed an accelerated narrowing of the neural folds followed by a significant widening of the distance between the folds (Fig. 1E-G). These data demonstrate that zinc deficiency alters inflection. There was a relatively weak effect of zinc deficiency on zippering from the caudal zipper point (Fig. 1H), but the caudal zipper point angle changed approximately coincident with the alteration in inflection (Fig. 1I). One obvious feature of the 15 µM TPEN-treated embryos is the reduction in the brain size and this was mitigated by the addition of zinc (Fig. 1C,D,J). Taken together, the major effect of TPEN-induced zinc deficiency is an alteration in neural fold inflection and reduced brain size.

The time-lapse videos of 15 µM TPEN-treated embryos show significant amounts of apparently dead tissue coming out from the open neural tube and the rostral-caudal size of the brain is significantly reduced 6 h after TPEN treatment. This suggests abnormal apoptosis as the major cause of the zinc deficiency-induced NTD at the cellular level. Using the lower concentration of TPEN (10 µM) and a shorter culture time (30 min and 2-4 h), we observed increased TUNEL-positive and cleaved caspase 3 (Casp3)-positive cells in the cranial regions of 6- to 10-somite stage embryos at the start of culture (Fig. 2A,B), and this could be suppressed by addition of zinc (Fig. 2C). The predominant tissue showing cleaved-Casp3 immunofluorescent staining was the neural ectoderm, with lesser and later staining in the non-neural ectoderm and even less in other regions, suggesting that the neuroectoderm is more sensitive to zinc deficiency than other cranial regions (Fig. 2A-C, Fig. S3). To determine whether a change in cell proliferation was also a contributor to the NTD, we assessed cell proliferation using phospho-histone H3 (pH3) staining and western blot analysis of 10 µM and 15 µM TPEN-treated E8.5 embryos. No changes were detected in the brain tissue (Fig. S4A,C). The lack of an observable change in cell proliferation, combined with the evidence that the cell cycle in the cranial neural tube is >12 h (Cameron and Greulich, 1963; Kim et al., 2007), suggests that the reduction in brain size and the NTD phenotype following TPEN treatment is due to excessive apoptosis. To evaluate the effect of zinc deficiency further, we turned to primary neuroepithelial cells isolated from E9.5 mouse brain. Consistent with the embryo results, TPEN treatment of neuroepithelial cells significantly increased the number of cleaved Casp3- and TUNEL-positive cells and zinc addition mitigated the TPEN-induced apoptosis phenotype, as shown by immunofluorescent staining and western blot analysis of cleaved Casp3 (Fig. 2D-F). This indicates that a major effect of zinc deficiency is excessive apoptosis during neural tube closure.

The Trp53 gene encodes the protein p53, a master regulator of cell survival. Increased levels of p53 initiate apoptosis, and p53 stability and activity are regulated by post-translational modifications and protein partner interactions. To determine whether zinc deficiency affects p53 levels, we treated E9.5 embryos for 3 h with TPEN and found that both cytoplasmic and nuclear levels of p53 in the brain region were dramatically increased; this induction was not observed upon culture with TPEN+ZnSO₄ (Fig. 3A). Similar results were found for primary E9.5 neuroepithelial cells (Fig. 3B). Immunofluorescent staining confirmed these findings of p53 stabilization in the nucleus and cytoplasm when zinc is deficient (Fig. 3C).

The NE-4C cell line is derived from p53 knockout E9 neuroepithelial cells (Livingstone et al., 1992; Schlett and Madarasz, 1997). To determine whether p53 is essential for zinc deficiency-induced apoptosis, we treated NE-4C cells with 10 µM TPEN for 1-3 h and found no difference in the number of cleaved Casp3-positive cells relative to control media (Fig. 3D,E). We then reintroduced p53 in NE-4C cells (see Materials and methods) and found that 1 µg of p53 plasmid increased the number of cleaved Casp3-positive cells, whereas 0.5 µg did not (Fig. 3D,E). Using the lower dose of p53 (0.5 µg), 10 µM TPEN treatment induced an increase in cleaved Casp3 positive cells over a 3 h treatment (Fig. 3D,E). Western blot for exogenous p53 again showed stabilization of p53 levels by TPEN treatment, which was decreased to baseline levels by addition of zinc (Fig. 3F). By contrast, TPEN treatment did not impact pH3 protein levels in NE-4C cells or brain tissue from E9.5 embryos, even when p53 levels were dramatically increased (Fig. S4B,C). These results indicate that zinc deficiency can elicit excess apoptosis through p53 stabilization.

p53 stability is determined in part by its ubiquitylation (Fig. 4A). The E3 ubiquitin ligase Mdm2 ubiquitylates p53, which drives it to the proteasome for degradation; conversely, the deubiquitylation enzyme Hausp removes ubiquitin and stabilizes p53 (Cheng et al., 2011; Haupt et al., 1997; Li et al., 2003; Li et al., 2002; Ma et al., 2010). We first examined the level of p53 ubiquitylation in primary E9.5 neuroepithelial cells treated with a protease inhibitor MG132 and 10 µM TPEN or TPEN+ZnSO₄, followed by p53 co-immunoprecipitation and probing with an anti-ubiquitin antibody. This revealed that the level of p53-associated ubiquitin signal was decreased after TPEN treatment, but addition of zinc recovered the levels of ubiquitylated p53 (Fig. 4B). As these experiments were performed in the presence of a proteasome inhibitor MG132, it suggests that p53 is not properly ubiquitylated upon zinc deficiency.

The N terminus of Mdm2 binds to p53 via the TAZ domain (Kussie et al., 1996), whereas its C terminus has a RING domain that requires zinc for proper assembly of Mdm2 and for its E3 ubiquitin ligase activity (Fig. 4A) (Fang et al., 2000; Nomura et al., 2017). Moreover, in some cells, Mdm2 can be cleaved by Casp3 at amino acid residue 359 or 361 to produce a short isoform of Mdm2 (∼60 kDa) that lacks the C-terminal RING domain (Erhardt et al., 1997; Pochampally et al., 1998). To determine whether zinc deficiency affects the interaction between p53 and Mdm2, we performed p53 co-immunoprecipitation assays on brain tissue from E9.5 embryos treated for 3 h with 10 µM TPEN or TPEN+Zn. This showed that binding of p53 to the full-length isoforms of Mdm2 (∼130 and ∼90 kDa) was significantly reduced by TPEN treatment, and this interaction could be restored by zinc supplementation (Fig. 4C,D). In contrast, binding of p53 to the short isoform of Mdm2 (∼55 kDa) was not affected by zinc deficiency, and the levels of total protein of the short isoform was dramatically increased (Fig. 4C,D). Similarly, in primary E9.5 neuroepithelial cells, 2 h after TPEN treatment, the binding of p53 to Mdm2 (∼90 kDa) was strongly diminished but p53 binding to the short isoform (∼55 kDa) was unchanged (Fig. 4E). The deubiquitylation enzyme Hausp also binds to ubiquitylated p53 and this binding was attenuated by TPEN treatment, whereas the addition of zinc increased p53 ubiquitylation and binding between p53 and Hausp (Fig. 4B-D). Taken together, these data suggest that zinc deficiency stabilizes p53 by disrupting Mdm2-mediated p53 ubiquitylation rather than increasing the interaction with Hausp deubiquitylation enzyme.

As a further test that the zinc deficiency-induced attenuation of p53 ubiquitylation allows a stabilization of p53 activity and leads to initiation of apoptosis, we manipulated the system with human CHIP (STUB1), a zinc-independent E3 ubiquitin ligase of p53 (Esser et al., 2005). CHIP overexpression in primary E9.5 neuroepithelial cells or in p53-transfected NE-4C cells can attenuate TPEN-induced p53 stabilization (Fig. S5A,B). Importantly, when CHIP was overexpressed in primary neuroepithelial cells, the TPEN-induced increase in cleaved Casp3-positive cells was significantly reversed (Fig. 4F,G). Taken together, we interpret these findings as indicating that zinc deficiency-induced p53 stabilization is due to reduced p53 ubiquitylation as a consequence of disruption of the zinc-coordinating RING domain of Mdm2, which leads to excess apoptosis during the process of neural tube closure.

p53 plays dual functions in the promotion of apoptosis: one as a transcription factor to regulate apoptosis-relevant target genes; the other by directly binding to anti-apoptotic Bcl2/Bcl-xl proteins to release pro-apoptotic mitochondrial Bax (Marchenko and Moll, 2007; Mihara et al., 2003; Toledo and Wahl, 2006; Vaseva and Moll, 2009). To attempt to discover the pathway affected by zinc deficiency-induced p53 stabilization during neural tube closure, we used two well-established pharmacological inhibitors of p53: pifithrin-α, an inhibitor of p53 transcriptional activity (Komarov et al., 1999; Sohn et al., 2009); and pifithrin-µ, which inhibits the p53 mitochondrial pathway and HSP70 (Sekihara et al., 2013; Strom et al., 2006). Time-lapse live imaging of 9-somite Ven^myr mouse embryos revealed that 40 µM pifithrin-α can clearly rescue TPEN-induced failure of neural tube closure (closure by 11 h±27 min; n=6; Fig. 5A,B, Movie 4). In contrast, pifithrin-µ (50 µM) failed to promote neural tube closure (n=5; Fig. 5A,B, Movie 5). These results suggest that p53 transcriptional activity is the main pathway deployed in the context of zinc deficiency, leading to NTD. The features of neural fold inflection were significantly improved by pifithrin-α co-treatment with TPEN and became similar to control (Fig. 5C-F). Moreover, pifithrin-α reduced the TPEN-induced increase in cleaved Casp3-positive cells in the midbrain and hindbrain regions in 9- to 10-somite embryos (Fig. 5G). Analysis of gene expression changes in primary E9.5 neuroepithelial cells showed that TPEN increased the expression of p53 target genes: the pro-apoptotic genes Fas1 (Fas) and Bax, as well as Mdm2 which is transcriptionally regulated by p53 in a negative feedback loop (Fig. 5H-J), but not the Cdkn1a gene, which encodes a cell cycle arrest protein p21^Cip1 that is targeted by p53 (Fig. 5K) (el-Deiry et al., 1993). These p53 target pro-apoptotic genes were returned to normal levels by TPEN+ZnSO₄ treatment (Fig. 5H-J). These data indicate that p53 transcriptional activity predominates in the promotion of apoptosis when zinc is deficient, resulting in failure of neural tube closure.

Epidemiological data in humans has implicated zinc deficiency as a risk factor for NTDs. Here, we sought to develop a mouse model of zinc deficiency to explore the cellular and molecular basis for failure of neural tube closure. A distinct advantage of our system, in which zinc levels are manipulated acutely, is that we can evaluate the immediate direct response, versus potential indirect consequences. We demonstrate that acute zinc deficiency in embryos causes excessive cell death, starting within 1 h and increasing thereafter, and ultimately results in failure of neural tube closure. Moreover, using live imaging, we find that neural fold zipping is not grossly affected but inflection of the neural folds towards each other is disrupted, preventing the neural folds from meeting at the midline. The early embryonic neuroepithelium appears most sensitive to zinc deficiency. Consistent with our findings of neural cell death, others have shown that reduced zinc conditions lead to cell death of neuroepithelial cells in embryos from wild-type dams given a zinc-deficient diet (Dufner-Beattie et al., 2006; Harding et al., 1988), in neuroblastoma and neural progenitor-like cells (Adamo et al., 2010; Nuttall et al., 2015; Seth et al., 2015), in adult neural stem cells (Corniola et al., 2008), and in cortical neurons (Adamo et al., 2010; Ra et al., 2009). Other effects of zinc deficiency include: reduced expression of the neural stem cell marker nestin in embryos and pups from dams fed a severely zinc-deficient diet (Wang et al., 2001); altered neural differentiation of neural precursor cells induced to differentiate with retinoic acid (Morris and Levenson, 2013) and of human induced pluripotent stem cells differentiated to motor neurons (Pfaender et al., 2016); and decreased fetal (rat E19) neural progenitor cell proliferation (Nuttall et al., 2015). Adult mice fed a zinc-deficient diet for 5 weeks show a decrease in proliferating neural cells and immature neurons, and apoptosis in the hippocampus (Gao et al., 2009). However, our acute embryonic zinc-deficiency studies did not detect a change in cell proliferation, although it is possible that longer periods of zinc deficiency may disrupt proliferation of these early neural progenitor cells. Zinc dyshomeostasis, either an excess or deficiency, is also implicated as a risk factor for neurodegenerative disorders (Szewczyk, 2013). Together, the data highlight zinc homeostasis as being crucial for neural development from embryonic through adult neurogenesis. Our in vivo and in vitro results of the earliest neural progenitor cells implicate apoptosis as the predominant phenotype associated with zinc deficiency-induced NTDs.

Our results that include rescue experiments point to p53 as being an important mediator of the apoptotic phenotype, although we are aware that the full response to zinc deficiency is likely molecularly complex and broader than the studies here have covered. Indeed, others have suggested that zinc deficiency alters binding activity of transcription factors involved in neuronal differentiation (Gower-Winter et al., 2013; Morris and Levenson, 2013), and decreases proliferation by reducing ERK1/2 signaling (Nuttall et al., 2015).

Here, we provide another mechanism related to p53 stabilization, through disruption of Mdm2 activity, for consideration of zinc function in neuroepithelial progenitor cells. Earlier studies found that p53 is stabilized when zinc is deficient (Ra et al., 2009). Our studies support this finding and extend it to an in vivo context and add mechanistic insight. In the developing mouse embryo, we found that acute zinc deficiency stabilizes p53 through disruption of its ubiquitylation. This increases p53 levels and causes excessive apoptosis and NTD, consistent with the role of p53 in regulating the response to cellular stress and initiating apoptosis.

p53 protein levels in embryonic neural tissue and primary neuroepithelial cells increase within 2 h of TPEN-induced zinc deficiency. This relatively rapid response suggests that p53 accumulation is due to inhibition of p53 degradation. p53 degradation is mediated by ubiquitylation of p53, which targets it to the proteasome. We find that p53 ubiquitylation is greatly reduced upon zinc deficiency. Mdm2 is the major E3 ubiquitin ligase for p53 and Mdm2 ubiquitylation activity is zinc dependent (Fang et al., 2000; Nomura et al., 2017; Shloush et al., 2011). Our results, in combination with biochemical studies of others, suggest that Mdm2 E3 ligase function is disrupted as a result of attenuated Mdm2 assembly and Mdm2 RING domain activity upon zinc deficiency. We show that zinc deficiency reduced binding of full-length Mdm2 to p53, and resulted in cleavage of Mdm2 to a shorter isoform without E3 ligase activity. It is likely that the culmination of these zinc-deficiency mediated changes to Mdm2 is responsible for the reduction in p53 ubiquitylation that results in p53 stabilization.

Zinc deficiency results in increased p53 in the cytoplasm and in the nucleus. We postulate that the increase in the nucleus is from two sources. First, translocation of stabilized p53 from the cytoplasm to the nucleus, consistent with results in which zinc-deficient conditions cause p53 translocation from the cytoplasm to the nucleus (Corniola et al., 2008). Second, inhibition of p53 nuclear export, as this is mediated by the Mdm2 zinc-coordinating RING-finger domain (Geyer et al., 2000). Zinc deficiency decreases the levels of the full-length RING-finger domain containing isoform of Mdm2 and attenuates binding of p53 to this isoform. Therefore, zinc deficiency would be predicted to decrease p53 nuclear export, and indeed we find an increase of p53 in the nucleus upon TPEN treatment of E9.5 neuroepithelial cells both in vivo and in vitro (Fig. 3A-C).

Consistent with an increase of p53 in the nucleus upon TPEN treatment, we also observed increased expression of p53 target genes (Fig. 5H-J). To distinguish between zinc deficiency-induced p53-mediated apoptosis due to p53 transcriptional activity or mitochondrial pro-apoptotic function, we used pharmacological inhibitors specific to each of these pathways. Although we cannot rule out a possible role for the mitochondrial pathway, our results clearly show that inhibition of p53 transcription activity is sufficient to prevent the TPEN-induced NTD. We also observe a robust increase of p53 in the nucleus upon TPEN treatment (Fig. 3A-C), where it can then act as a transcription factor. Our studies show zinc deficiency induces the expression of p53 pro-apoptotic target genes Bax and Fas1, as well as Mdm2, which is autoregulated by p53, consistent with previous studies (Gao et al., 2009; Seth et al., 2015). Furthermore, the ability of the p53 transcriptional inhibitor to relieve zinc deficiency-induced apoptosis and to restore neural tube closure provides evidence that increased p53 transcriptional activity, due to disruption of Mdm2-regulated p53 stabilization and nuclear accumulation, is responsible for the zinc deficiency-induced NTDs.

Zinc is indispensable for multiple molecular events relevant to p53. The DNA-binding domain (DBD) of p53 contains zinc and the level of zinc ions is crucial for proper DBD folding and p53 DNA-binding activity (Butler and Loh, 2003, 2007; Duan and Nilsson, 2006). This might predict that zinc deficiency would decrease p53 DNA-binding activity and lead to a decrease in apoptosis. However, our in vivo data over a relatively short timescale of zinc deficiency indicate that p53 target gene expression is enhanced and apoptosis is increased, suggesting that at least the acute effect of zinc deficiency in the neuroepithelium is not mediated via an effect on the p53 DBD.

Our model of at least one predominant effect of zinc on neural tube closure is schematized in Fig. 6. During normal neural tube closure, when zinc levels are adequate, the p53-Mdm2 relationship is intact, allowing Mdm2 to modulate p53 levels to maintain cell survival and inhibit cell death. Our study highlights a mechanistic pathway in which zinc deficiency leads to failure of neural tube closure through attenuation of p53 ubiquitylation by Mdm2, p53 stabilization, increased p53 nuclear accumulation, enhancement of p53 transcriptional activity and initiation of extensive cell death, which prevent the neuroepithelial folds from continuing their progress towards the dorsal midline to meet and form a closed neural tube.

C57BL/6J mice were used for E9.5 brain tissue collection and primary E9.5 neuroepithelial cell isolation. Transgenic mice CAG::myr-venus (Rhee et al., 2006) [Jackson Labs Tg(CAG-Venus)^1Hadj/J] were used for time-lapse live imaging data collection. All mice were housed at the University of Colorado Anschutz Medical Campus in accordance with an IACUC approved protocol.

As described in our previous publications (Massarwa and Niswander, 2013; Ray and Niswander, 2016), somite-staged embryos were placed in culture media on a mounting platform for imaging on a Zeiss LSM 880 Meta confocal microscope on a motorized stage with a temperature- and gas-controlled incubator (37°C, 5% O₂, 5% CO₂). Culture medium was DMEM without Phenol Red and contained 55 U/ml penicillin/streptomycin, 2.2 mM glutamax, 11 mM Hepes buffer, mixed with 75% rat serum (Harlan, #B-4520). The embryos were mounted on black Millipore filter paper (AABG01300). Glass-bottomed dishes were from MatTek (P35G-1.5-20-C). Images were acquired using a 10× lens c-Apochromat NA1.2 with 0.7 digital zoom out. A 448 nm laser (around 2% power) was used. Images were acquired at 512×512 resolution. The thickness of z-slices was 15 μm (a total of 15-30 slices/time point, depending on the experiment). The time interval for imaging was ∼5 min. Post-imaging processing and measurements were carried out using the LSM image browser (Zeiss 880). The movies were exported using Imaris 8.0 software.

For molecular studies, E8.5 or E9.5 embryos were dissected with intact yolk sac and placed in a roller culture system similar to that used by Pyrgaki et al. (2010) in a gas-controlled incubator (37°C, 5% O₂, 5% CO₂). The same culture medium as above was mixed with 50% rat serum. Embryos were treated for 30 min to 4 h in Fig. 2A,B, for 4 h in Fig. 2C, for 3 h in Fig. 3A and for 3 h for experiments in Fig. 4C,D. For western blot and co-IP assays, brain tissues from four or five embryos exposed to the same treatment were pooled for protein extraction as one sample.

E9.5 neuroepithelial cells were dissected from whole-brain tissue of E9.5 mouse embryos and pipetted up and down to dissociate cells. The culture medium was as previously published for E8.5 neuroepithelial cells (Tropepe et al., 1999) using a chemically defined serum-free basic medium composed of a 1:1 mixture of Dulbecco's modified Eagle's medium (DMEM; 11995, GIBCO) and F-12 nutrient (11765-054, GIBCO) including 0.6% glucose (D9434, Sigma), 2 mM glutamine (25030081, GIBCO), 3 mM sodium bicarbonate (25080, GIBCO), 5 mM HEPES buffer (15630, GIBCO), 25 µg/ml insulin (I9278, Sigma), 100 µg/ml transferrin (T8158, Sigma), 20 nM progesterone (P6149, Sigma), 60 µM putrescine (P5780, Sigma) and 30 nM selenium chloride (323527, Sigma). FGF2 (10 ng/ml) (SRP4037, Sigma), 2 µg/ml heparin (H3149, Sigma) and 20 ng/ml EGF (62253-63-8, Sigma) were added into the basic medium as a growth medium. NE-4C cells (ATCC #CRL-2925) were grown at 37°C and 5% CO₂ in MEM (Invitrogen) supplemented with 5% FBS, 1×MEM nonessential amino acids (Invitrogen) and 1×GlutaMax (Invitrogen). The NE-4C cell line has been tested in the lab for mycoplasma contamination.

To establish a zinc-deficient environment, N,N,N′,N′-Tetrakis (2-pyridylmethyl) ethylenediamine (TPEN) (P4413, Sigma) was dissolved in DMSO and used at a final concentration of 10 µM or 15 µM to treat cells or embryos according to the previous studies in which 1-100 µM were employed (Gao et al., 2017; Rayman et al., 2018; Strand et al., 2013). For zinc supplementation, zinc sulfate heptahydrate (Z4750, Sigma) was dissolved in water and used at a final concentration of 25 µM or 50 µM ZnSO₄ (Gao et al., 2017). Pifithrin-α (1267, Tocris) and pifithirin-µ (2653, Tocris) were used to inhibit p53 transcriptional activity and the mitochondrial pro-apoptotic pathway, respectively. Pifithrin-α and pifithrin-µ were dissolved in DMSO and used at a final concentration of 40 µM for pifithrin-α and 50 µM for pifithrin-µ according to previous studies in which 3-100 µM pifithrin-α (Komarov et al., 1999; Sohn et al., 2009) and 10-50 µM pifithrin-µ were employed (Strom et al., 2006).

The cytoplasm and nuclei were fractioned using a Nuclear Complex Co-IP kit (54001, Active Motif). Proteins extracts were subjected to polyacrylamide gel electrophoresis and electrotransfer to PVDF membranes. The membranes were blocked with 5% non-fat milk and then probed with the indicated antibodies in TBS-Tween 20. Antibodies used were against p53 (1C12) (1:500; 2524; Cell Signaling Technology); cleaved caspase 3 (Asp175) (1:500; 9661; Cell Signaling Technology); Hausp (D17C6) (1:500; 4833; Cell Signaling Technology); Mdm2 (SMP14) (1:50; sc-965; Santa Cruz); Chip (C3B6) (1:500; 2080; Cell Signaling Technology); ubiquitin (1:500; 3933; Cell Signaling Technology); DDK (FLAG OTI4C5) (1:2000; TA50011; Origene); pH3 (1:500; 9714; Cell Signaling Technology) H3 (1:14,000; ab1791; Abcam); and Gapdh (Sigma; G9545; 1:2000). ImageJ was used to quantify the western blot data. Integrated gray value was measured. In all assays, relative protein level was standardized by dividing the referenced protein (usually Gapdh in cytoplasm or H3 in the nucleus), and then one treatment was designated as 1, and the other treatments were normalized to this treatment.

Freshly dissected embryonic brain tissue or E9.5 neuroepithelial cells were collected, and total protein was harvested. Anti-p53 (1C12) antibody (2524; Cell Signaling Technology) (5μl) was added to 500 μg of input and rotated overnight at 4°C. Washed PureProteome Protein A/G Mix Magnetic Beads (50 μl; LSKMAGAG02; Millipore) were added and samples incubated at 4°C for 1 h on a rotator. Beads were washed five times in ice-cold wash buffer before elution in 2× Laemmli buffer at 95°C for 10 min. Protein interactions were subsequently assayed via western blot and ImageJ was used to quantify the data. In all assays, relative protein level was standardized by dividing the p53 protein level in immunoprecipitated protein, and then one treatment was designated as 1, the other treatments were normalized to this treatment.

To reintroduce p53 or overexpress CHIP protein, NE-4C cells or E9.5 primary neuroepithelial cells were subjected to transient transfection. Plasmids containing the coding region of human p53 (TP53-Myc-DDK-tagged) (RC200003, Origene) or human CHIP protein (STUB1-Myc-DDK-tagged) (RC200310, Origene) were transfected into cells using Xfect Transfection Reagent (Clontech) according to the manufacturer's protocol (1.5 µg or 2.5 µg plasmid for transfection per well of a 12-well plate; 0.5 or 1 µg for 24-well plate). Approximately 24 h after transfection, 15 µM TPEN was added and the cultures collected 1-3 h later for immunofluorescent staining or western blot analysis.

Embryos or cultured cells were fixed for 10-15 min at room temperature in 4% PFA, blocked for 1 h at room temperature, and stained with primary antibody overnight at 4°C. Primary antibodies used were as follows: p53 (1C12) (1:300; 2524; Cell Signaling Technology); cleaved caspase 3 (Asp175) (1:300; 9661; Cell Signaling Technology); nestin (SP103) (1:100; ab105389; Abcam); pH3 (1:300; 9714; Cell Signaling Technology). Hoechst (1:1000)-labeled nuclei were counted as a cell number control.

In Situ Cell Death Detection Kit (12156792910; Roche) was used for TUNEL assays. Embryos or cultured cells were fixed by 4% paraformaldehyde for 10-15 min at room temperature, then a TUNEL-labeled assay was performed following the manufacturer's instructions.

Imaging was performed on a Zeiss LSM880 Meta laser scanning confocal microscope using Zen software. Imaris 8.0 software was used for cell counting quantitative analysis. The number of Hoechst-labeled cells in a field were counted and the percentage of cleaved-Casp3 or TUNEL-positive cells quantified. In embryos, the number of cleaved Casp3-positive, TUNEL-positive or pH3-positive cells were counted in hindbrain and midbrain regions.

E9.5 primary neuroepithelial cells were harvested 6 h after 15 µM TPEN or 15 µM TPEN+ 25 µM ZnSO₄, or 25 µM ZnSO₄ treatment. RNA was isolated according to manufacturer's instructions using the High Pure RNA Isolation Kit (Roche) and cDNA was generated using random hexamers and the Transcriptor First Strand cDNA Synthesis Kit (Roche). qPCR was performed on a LightCycler 480 (Roche) using the LightCycler 480 Probes Master reagent (Roche) and Universal Probe Library (Roche) according to the manufacturer's protocol. All primers and probes are listed in Table S1. Data were collected and analyzed with LightCycler 480 Software (Roche; Version 1.5.1). Actb gene is referenced gene. All experiments were from three biological replicates.

One-way ANOVA was used to analyze the experimental data, post-hoc Tukey HSD multiple comparisons were performed in multiple treatment assays. Values are represented as mean±s.d.; P<0.05 is considered as statistically significant.

We thank our lab for constructive advice and James Orth for critical reading of the manuscript.