Abstract
Prenatal alcohol exposure is a leading cause of permanent neurodevelopmental disability and can feature distinctive craniofacial deficits that partly originate from the apoptotic deletion of craniofacial progenitors, a stem cell lineage called the neural crest (NC). We recently demonstrated that alcohol causes nucleolar stress in NC through its suppression of ribosome biogenesis (RBG) and this suppression is causative in their p53/MDM2-mediated apoptosis. Here, we show that this nucleolar stress originates from alcohol’s activation of AMPK, which suppresses TORC1 and the p70/S6K-mediated stimulation of RBG. Alcohol-exposed cells of the pluripotent, primary cranial NC line O9-1 were evaluated with respect to their p70/S6K, TORC1, and AMPK activity. The functional impact of these signals with respect to RBG, p53, and apoptosis were assessed using gain-of-function constructs and small molecule mediators. Alcohol rapidly (<2hr) increased pAMPK, pTSC2, pRaptor, p-mTOR(S2446), and reduced both total and p-p70/S6K in NC cells. These changes persisted for at least 12hr to 18hr following alcohol exposure. Attenuation of these signals via gain- or loss-of-function approaches that targeted AMPK, p70/S6K, or TORC1 prevented alcohol’s suppression of rRNA synthesis and the induction of p53-stimulated apoptosis. We conclude that alcohol induces ribosome dysbiogenesis and activates their p53/MDM2-mediated apoptosis via its activation of pAMPK, which in turn activates TSC2 and Raptor to suppress the TORC1-p70/S6K-mediated promotion of ribosome biogenesis. This represents a novel mechanism underlying alcohol’s neurotoxicity and is consistent with findings that TORC1-p70/S6K networks are critical for cranial NC survival.
1. Introduction
Prenatal alcohol exposure (PAE) is a leading cause of permanent neurodevelopmental disability and features deficits in cognition and executive function [1]. Diagnosis is often initiated by a distinctive craniofacial appearance [2] that has complex origins reflecting both changes in the underlying brain growth [3,4] and direct effects upon the cranial neural crest [5,6], a pluripotent stem cell linage that forms the facial bone and cartilage, and neuronal elements including certain cranial nerves, Schwann cells, and the sympathetic and parasympathetic system [7]. Neural crest induction begins at neurulation and, shortly thereafter, they emigrate from the closing dorsal neural tube and into peripheral tissues where they differentiate into the aforesaid structures. Pharmacologically relevant alcohol exposure (10-80mM) disrupts multiple events in cranial neural crest development including their induction, migration, proliferation, and survival [4–6]. We and many others have shown that alcohol causes the apoptotic deletion of neural crest progenitors at a specific phase of their development, at the onset of migration [8–10]. Alcohol interacts with multiple pathways to modulate facial outcome and/or neural crest death including sonic hedgehog, retinoid, oxidative stress, Wnt, and p53, among others [11–17]. However, the regulatory mechanisms governing neural crest sensitivity to alcohol remain elusive. Novel insight into this apoptotic mechanism emerged from our whole exome sequencing of neural crest progenitors, wherein a brief alcohol exposure (2hr, 52mM) rapidly (<6hr) repressed the expression (padj=10E-47) of 106 nuclear and mitochondrial ribosomal proteins (RP) by 30%-50% [18]. A similar repression of RPs is observed in alcohol-exposed zebrafish [19], headfolds of mouse, [20,21] and chick strains [22] having differential alcohol vulnerability, primary mouse neural stem cells [23] and in the cranial neural crest primary cell line O9-1 [24]. Thus, suppression of RP synthesis is a consistent feature of alcohol-exposed neural crest and neuroprogenitors across taxons.
Why would a suppression of RPs lead to the apoptosis of cranial neural crest populations? For rapidly proliferating populations such as neural crest, each cell division requires the duplication not only of DNA, but the ribosomes that accomplish protein synthesis. Indeed, the production of rRNA plus the ∼200 proteins and short RNAs involved in rRNA processing and assembly into ribosomes is estimated to occupy 70-80% of the cellular energy budget [25], and is seen visually in the nucleosomes, which are the multiple operon sites of this ribosome biogenesis (RBG) [26]. Unsurprisingly, cells have coopted RBG to monitor their internal stress, and RBG is tightly coupled to cellular anabolism and the p53 checkpoint pathway [27,28]. Specifically, under normal conditions RBG is stimulated by the anabolic effector TORC1 via p70/S6K [28]. When energy is limiting, pAMPK phosphorylates the TORC1 components TSC1/2 and Raptor to suppress TORC1 and RBG, thus integrating RBG with cell metabolism. Under stressors that reduce the resources needed for rRNA synthesis, RBG ceases and the nucleoli visually disappear [29,30]. This process, known as “nucleolar stress”, is linked to p53 via the nuclear E3 ubiquitinase MDM2, which under normal conditions targets and destabilizes p53 [31–33]. Under nucleolar stress, reduced rRNA synthesis enables a RPS5-RPL11-5SrRNA complex to bind and inhibit MDM2, thus allowing p53 to become transcriptionally active [31–33]. The importance of RBG for neural crest is evidenced in the genetic disorders known as ribosomopathies, in which loss-of-function in AMPK-TORC1-p70/S6K signaling, or in components of RBG such as rRNA, RPs, or the ribosome assembly machinery, cause p53-mediated neural crest losses and facial deficits that share similarities with those of PAE (i.e. Treacher-Collins syndrome, Diamond-Blackfan anemia) [34–36].
We recently reported that alcohol exposure impairs RBG to induce nucleolar stress in early cranial neural crest progenitors. Alcohol exposure causes the rapid cessation of rRNA synthesis and dissolution of nucleolar structures in response to pharmacologically relevant alcohol exposures (20-80mM for 2hr) [24]. This is followed by the stabilization of nuclear p53 and their subsequent p53-mediated apoptosis [17,24]. Haploinsufficiency of RPL5, RPL11, or RPS3A using a morpholino approach synergized with alcohol to cause craniofacial deficits in a zebrafish model of PAE [18,24], and transfection with MDM2 or loss-of-function p53 blocked these effects of alcohol [24]. However, the mechanism responsible for alcohol’s suppression of RBG is unknown. Here we test the hypothesis that alcohol represses RBG through its suppression of p70/S6K via TORC1, and that its activation of AMPK mediates both the loss of TORC1-p70/S6K signaling and their p53-mediated apoptosis.
2. Materials and Methods
2.1. Cell Culture
All studies use the established primary cranial NC line O9-1 (#SCC049, Millipore; Burlington, MA) [37,38]. This non-transformed stem cell line was originally isolated from mass culture of primary cranial NCs isolated at embryonic day (E)8.5 from C57BL/6J mouse embryos that expressed Wnt1-Cre:R26R-GFP [37]. It expresses key NC markers (e.g., Twist1, Snail1, Nestin, CD44), can differentiate into osteoblasts, chondrocytes, smooth muscle cells, tooth pulp, and glial cells, but not neuronal cells, and develop normally when added back to embryos [37–40]. We find it has alcohol responses identical to those of primary avian or mouse NCs [24]. Cells were maintained on Matrigel-coated plates in their pluripotent state at 50-60% confluence in DMEM (Gibco, Grand Island NY) supplemented with 6% ES cell media (ES-101B, Sigma, Louis MO), 15% fetal calf serum (Millipore), 0.1mM nonessential amino acids (Gibco), 1000 U/ml leukemia inhibitory factor (LIF, Gibco), 25ng/ml basic fibroblast growth factor (Invitrogen, Waltham, MA), 0.3µM β-mercaptoethanol (MP Biomedical, Solon, OH), and 100 U/ml penicillin and streptomycin (Gibco). Experiments were performed in the identical media but containing 1% FCS conditioned media and were tested within 18hr.
2.2. Alcohol Exposure
Cells were treated with USP grade ethanol (Koptec, King of Prussia, PA) at 80mM; we previously found this is the EC50 for the alcohol-mediated dissolution of nucleolar structures in O9-1 cells [24]. In these culture conditions, we find that the alcohol concentrations drop to less than 10mM by 2hr post-addition. Thus, these studies model acute binge exposure.
2.3. Quantitative PCR
RNA was isolated using Trizol reagent (Invitrogen) at experimentally determined times following exposure to 80mM alcohol. Total RNA (1µg) was reverse transcribed with 500µg/ml random primer (#C1181), 5X ImProm II reaction buffer (#M289A), ImProm II (#M314A), 25 mM MgCl2 (#A351H), 10 mM dNTP (#U1511), 20U RNAsin (#N2511; all from Promega, Madison, WI) and RNAse free water. Quantitative PCR (qPCR) were performed using SYBR Select Master Mix (ABI, #4472913) and the Real-Time PCR system (Bio-Rad CFX96; Hercules, CA). De novo rRNA synthesis was quantified by qPCR for the Internal Transcribed Spacer ITS1 (138bp) located between the 18S and 5.8S rRNA of newly transcribed 47S rRNA [23,24]. The primers (IDT, Coralville, IA) for ITS-1 were forward: 5′-CCGGCTTGCCCGATTT-3′, reverse: 5′-GCCAGCAGGAACGAAACG-3′ and for β2 microglobulin were forward: 5′-TTCACCCCCACTGAGACTGAT-3′; reverse: 5′-GTCTTGGGCTCGGCCATA-3′. Relative abundance was calculated using the 2- ΔΔCT method and was normalized to β2 microglobulin. qPCR methodology adhered to the MIQE standards [41].
2.4. Western Immunoblot Analysis
Total proteins were isolated from cells using NP-40 lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% NP-40, 50 mM NaF, 1 mM NaVO3, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride; all from Sigma). Total protein was separated on 10% SDS-PAGE reducing gels for proteins ≤150kDa, or 4%-20% detergent gels for those ≥150kDa (BioRad). The resolved proteins were transferred to a PVDF membrane using the Trans-Blot Turbo Transfer System (Bio-Rad). Primary antibodies directed against p-AMPK(Thr172) (#2535), AMPK (#2793), p-mTOR(Ser2446/2448) (#2971), mTOR (#4517), p-Raptor(Ser792) (#2083), Raptor (#2280), p-p70/S6K1(Thr389) (#9234), p70/S6K1 (#9202), pRictor(Thr1135) (#3806), Rictor (#2140), pTSC2(Ser1387) (#5584), TSC2 (#4308), PKCα (#2056), p4E-BP1(Thr70) (#9455), and 4E-BP1(#9452) were all used at 1:1000 and were from Cell Signaling (Danvers MA). Antibodies directed against p53 (#ab26, 1:500), Deptor (#191841, 1:200) and pPKCα(Ser657) (#ab180848) were from Abcam (Cambridge, England, UK). Anti-GAPDH (#G8795, 1:2500) was from Sigma. The secondary antibodies were goat anti-rabbit (#4010-05, 1:5000) and anti-mouse immunoglobulin (#1030-05, 1:5000, both from Southern Biotech, Birmingham AL) coupled to horseradish peroxidase. Signal was imaged by chemiluminescence using the Radiance Q detection system (Azure Biosystems, Dublin, CA).
2.5. Cell Transfection
Cells were transiently transfected at 80% confluence with plasmid pcDNA3 p70/S6K2 E388 D3E (#17731; Addgene; Watertown, MA; 42) and Lipo-3000 (#L3000001; Invitrogen) according to the manufacturer’s instructions. This construct contains four gain-of-function mutations (T388E, S410D, S417D, S423D) that confer rapamycin-resistance and constitual activity (ca-S6K; 42). Transfected cells recovered for 6hr prior to starting the alcohol exposure.
2.6. Apoptosis Analysis
Cells were treated with 80 mM alcohol. At 16hr thereafter they were fixed in 4% paraformaldehyde, permeabilized in 0.2% TritonX-100, and apoptosis was detected using TUNEL assay (In Situ Cell Death Detection kit, TMR Red; Roche, Indianapolis, IN). Nuclei were visualized using DAPI. Images were taken under uniform exposure. Data are the means of three independent experiments, each having three samples per group and quantifying three images per sample.
2.7. Small Molecule Interventions
These used the AMPK inhibitor dorsomorphin (5µM), the AMPK activator AICAR (500µM), the TORC1 inhibitor rapamycin (1 nM; all from Selleck Chemicals, Houston TX), or the TORC1 activator L-leucine (250 µM, Sigma). For all, cells were exposed for 1h and then treated with 80mM alcohol and collected at experimentally determined times thereafter.
2.8. Statistical Analysis
Data were tested for normality and then analyzed using the appropriate test, Student’s t-test for comparisons of two groups and one-way ANOVA for multiple comparisons. P < 0.05 was considered significant. Unless otherwise indicated, all results are presented as the mean ± SD of 3 independent samples per group.
3. Results
3.1. Alcohol suppresses RBG and induces nucleolar stress and p53-mediated apoptosis in neural crest
As we showed previously [24], a time course reveals that exposure to 80mM alcohol suppressed de novo rRNA synthesis by 1hr following exposure (CON 100 ± 11.7%, ALC 63.5 ± 4.8%; p<0.001; Figure 1A), and this reduction persisted for at least 4hr to 8hr thereafter. This was followed by a dissolution of nucleolar structures that was visualized using antibodies directed against the RNA polymerase I transcription factor UBF (Figure 1B) [24,29]. The proapoptotic protein p53, which responds to nucleolar stress via MDM2 [31–33] was stabilized in these cells within 2hr of exposure, when its content increased 70 ± 11% (p<0.001). We showed previously [37] that this largely represents nuclear protein and its elevation persisted for at least 12hr (CON 100 ± 42%, ALC 191 ± 14%; p<0.001; Figure 1C), whether normalized to GAPDH (upper blots) or total protein (lower blot). By 16hr, the alcohol-exposed NC underwent apoptosis (CON 0.97 ± 0.23%, ALC 10.94 ± 1.24%; p<0.001; Figure 1D), and we showed previously that, as in other models [16,17,23], this apoptosis is p53-dependent because it is prevented by transfection with dominant-negative p53 [24].
3.2. Alcohol exposure reduces p70/S6K and S6K gain-of-function normalizes rRNA synthesis and apoptosis in neural crest
To gain insight into the mechanism by which alcohol induced nucleolar stress in neural crest, we evaluated its impact upon a primary effector of RBG, p70/S6K [28]. Alcohol exposure reduced the abundance of p70/S6K in a complex manner. With respect to total p70/S6K protein, its content was 69 ± 6% (p<0.001) of CON values within 2hr of alcohol exposure, and these reductions persisted through 6hr and were normalized by 12hr post-exposure (Figure 2A). This reduction was mediated, in part, at the transcriptional level because the abundance of transcripts encoding p70/S6K was 89.7% of CON (padj=4.28E-12) as determined by whole exome sequencing. With respect to its activated form, phospho-p70/S6K-Threonine(T)389, its levels did not decline as a percentage of total p70/S6K until 12hr post-exposure, when they dropped to 57 ± 1% at 12hr (p<0.001) and 59 ± 7% at 18hr (p=0.004) post-exposure. However, because of the absolute decline in p70/S6K, this represented a 30-40% decline in the abundance of activated p70/S6K within 2hr of the alcohol exposure.
To test the functional involvement of this p70/S6K loss, we transfected NC with a constitutively active (ca) version of p70/S6K (caS6K), which contains four gain-of-function mutations that confer rapamycin-resistance (42). As expected, the caS6K construct countered the alcohol-mediated reduction in p-p70/S6K(T389) (ALC, 55 ± 7%, ALC+caS6K, 83 ± 12%, p=0.022; Figure 2B) to levels not different from CON (1.00 ± 0.16, p>0.10). It restored de novo rRNA synthesis in alcohol-exposed cells (ALC 49 ± 5%, ALC+caS6K 90 ± 3%, p<0.001; Figure 2C), abrogated alcohol’s activation of p53 (ALC, 155 ± 29%, ALC+caS6K, 106 ± 14%; p=0.05; Figure 2D), and prevented their apoptosis (Figure 2E, ALC, 2.70 ± 0.45%, ALC+caS6K, 0.52± 0.11%; p <0.001). In the controls, caS6K unexpectedly reduced rRNA synthesis (CON, 100 ± 6%, CON+caS6K, 54 ± 1%, p<0.001), but this had no effect on p53 content or apoptosis. We conclude that alcohol rapidly suppressed p70/S6K and that this contributes to the loss of RBG.
3.3. Alcohol activates the upstream TORC1 repressors pTSC2 and pRaptor to reduce TORC1 signaling activity
A primary activator of p70/S6K is the Target of Rapamycin Complex 1 (TORC1). TORC1 is under complex regulation via phosphorylation of its core component protein mTOR by kinases including Akt, AMPK, and p70/S6K itself [27]. Under alcohol exposure, we found increased phosphorylation of mTOR at the dual sites S2446/S2448 (Figure 3A); this increase was observed within 2hr of alcohol exposure and persisted for at least 12hr. Although these two sites could not be distinguished, they are under quite different regulation; S2446 is targeted by AMPK to reduce TORC1 activity, whereas S2448 is targeted by both Akt and p70/S6K to increase TORC1 signaling [27]; however, increased TORC1 signaling was inconsistent with the downstream reduction in p70/S6K. To disentangle whether this represented an activation or repression of TORC1, we evaluated TSC2 and Raptor, which act immediately upstream as AMPK-dependent negative regulators of TORC1 [43,44]. Alcohol increased the content of pTSC2(S1387) within 2hr of exposure (CON, 100 ± 1%, ALC. 144 ± 25%; p=0.05; Figure 3B) and increased p-Raptor(S792) by 6hr (CON, 100 ± 5%, ALC. 168 ± 29%; p=0.04); TSC2 was largely normalized by 6hr, whereas pRaptor remained elevated for at least 12hr (ALC 12hr, 143 ± 2%, p<0.001 vs CON). As further evidence that this represented a reduction in TORC1 activity, we found a reduction in a second TORC1 target p4E-BP1(T70) (CON, 111 ± 6%, ALC 85 ± 4%, p=0.027; Figure 3C), which is otherwise activated by TORC1 to increase protein translation in concert with p70/S6K’s stimulation of RBG [27,28]. We also explored alcohol’s impact on effectors of TORC2 and found no differences with respect to the abundance of pRictor(T1135), Deptor, pTSC2(S1387), or pPKCα(S657), suggesting alcohol’s suppression of TORC1 was selective in these neural crest cells (Figure 3D).
To explore these findings further, we tested if targeted inhibition of the TORC1 signaling pathway was sufficient to suppress RBG and induce NC apoptosis. Pretreatment with rapamycin (Rapa), a small molecule inhibitor of TORC1 [45], prior to alcohol exposure further suppressed rRNA synthesis (ALC, 80 ± 0.3%, ALC+Rapa, 59 ± 13%; p=0.03; Figure 4A), and aggravated p53 stabilization (ALC, 151 ± 21%, ALC+Rapa, 175 ± 8%; p=0.05; Figure 4B) and apoptosis (ALC, 5.19 ±0.37%, ALC+Rapa, 7.48± 0.76%; p=0.004; Figure 4C). Interestingly, rapamycin increased the p53 content in control cells but did not suppress their rRNA synthesis or cause apoptosis, indicating that TORC1 inhibition was insufficient, in of itself, to suppress RBG and induce apoptosis in these cells.
We further tested TORC1 contributions functionally by providing the TORC1 activator L-leucine just prior to the alcohol-exposure. We found that supplemental L-leucine (at 250% of normal cell culture levels) enhanced rRNA synthesis in alcohol-exposed cells (ALC, 77 ± 6%, ALC+Leu, 143 ± 26%, p=0.001; Figure 4D), and abrogated the loss of total p70/S6K protein (p=0.004) but did not further affect p-p70/S6K content (Figure 4E). It also normalized p53 levels (p<0.001; Figure 4F) and prevented their alcohol-induced apoptosis (p<0.001; Figure 4G). In CON, it increased p-p70/S6K levels (p=0.005) and rRNA synthesis (CON, 100 ± 5%, CON+Leu, 159 ± 15%, p=0.002). Taken together, these data suggested that acute alcohol rapidly repressed TORC1 activity and this might be mediated through the activation of TSC1 and Raptor.
3.4. Alcohol rapidly activates AMPK signaling in neural crest, and its inhibition prevents alcohol’s suppression of RBG and p53-mediated apoptosis
The canonical repressor of TORC1 activity is the AMP-dependent kinase (AMPK), which inhibits TORC1 via its phosphorylation of TSC2 and pRaptor [43,44]. Consistent with its actions upon TSC2, Raptor, TORC1, and p70/S6K, alcohol exposure caused a rapid elevation in the activated form of AMPK, pAMPK(Thr172), within 2hr of exposure (CON, 100 ± 1%, ALC 143 ± 6%, p<0.001; Figure 5A). The pAMPK content peaked between 6hr and 12hr (6hr, 188 ± 10%; 12hr, 184 ± 15%; p<0.001) and remained elevated through at least 18hr (150 ± 14%, p<0.001) following the alcohol exposure. To functionally test AMPK’s role in the loss of p70/S6K and RBG and this p53-mediated apoptosis, we treated cells with the AMPK inhibitor dorsomorphin prior to their alcohol exposure [43,44]. Dorsomorphin attenuated alcohol’s activation of pAMPK(T172) (ALC, 128 ± 7%, ALC+DOR, 89 ± 11%, p=0.005) and normalized the content of downstream targets pRaptor(S792) (ALC, 122 ± 8%, ALC±DOR, 27 ± 3, p<0.001), p-mTOR(S2246/8) (ALC 135 ± 6%, ALC+DOR, 69 ± 3%, p<0.001), and p-p70/S6K(T389) (ALC, 67 ± 6%, ALC+DOR, 135 ± 18%, p=0.007; Figure 5B). The AMPK inhibitor also normalized de novo rRNA synthesis (ALC, 70 ± 4%, ALC±DOR, 97 ± 6%, p<0.001; Figure 5C), and prevented the stabilization of p53 (ALC, 128 ± 5%, ALC±DOR, 58 ± 8%; p=0.003; Figure 5D) and subsequent apoptosis in response to alcohol (ALC 6.74 ± 0.92%, ALC+DOR 0.86 ± 0.06%; p<0.001; Figure 5E). These data suggested that alcohol caused neural crest apoptosis through the AMPK-mediated inhibition of TORC1 and p70/S6K to suppress their RBG.
We then asked if inappropriate activation of pAMPK was sufficient to invoke nucleolar stress and p53-mediated apoptosis in neural crest. Using the AMP mimetic and AMPK activator AICAR [43,44], we found that in otherwise normal cells it non-significantly elevated pAMPK(T172) (CON, 100 ± 6%, CON+AICAR, 127 ± 13%, p=0.094; Figure 6A) and reduced p-p70/S6K(T389) (CON, 100 ± 26%, CON+AICAR, 64 ± 31%, p>0.1; Figure 6B). However, it was not sufficient to suppress their rRNA synthesis nor did it stabilize p53 or cause their apoptosis (Figure 6C-E). In cells exposed to alcohol, it further elevated pAMPK(T172) (ALC, 133 ± 16%, ALC+AICAR, 172 ± 27%, p=0.001) and trended to further reduce p-p70/S6K(T389) content (ALC, 38 ± 16%, ALC+AICAR, 24 ± 4%, p=0.06). However, it did not further suppress rRNA synthesis, it modestly increased p53 (ALC, 120 ± 4%, ALC+AICAR, 140 ± 9%, p=0.03), and synergized with alcohol to worsen their apoptosis (ALC, 5.24 ± 0.50%, ALC+AICAR, 10.15 ± 0.51%, p<0.001). Taken together, these data indicate that alcohol’s activation of AMPK is responsible for the p70/S6K-mediated loss of RBG and activation of p53-mediated apoptosis in neural crest.
4. Discussion
The most important finding reported here is that alcohol exposure suppresses the TORC1-dependent activity of p70/S6K in cranial neural crest to inhibit ribosome biogenesis, induce nucleolar stress, and effect their p53-mediated apoptosis. Moreover, this suppression of p70/S6K was driven by alcohol’s upstream activation of pAMPK. Counteracting these changes through administration of pS6K gain-of-function or dorsomorphin-mediated inhibition of pAMPK were sufficient to sustain RBG in alcohol’s presence and thus prevent p53 stabilization and apoptosis. Alcohol’s action on this pathway was rapid, with a loss of rRNA synthesis within 1hr and increases in both pAMPK and p53 within 2hr post-exposure, suggesting these represent primary targets of alcohol. We reported previously that pharmacologically relevant alcohol exposures (20-80mM) suppress RBG and induce nucleolar stress in cranial neural crest cells, leading to their p53/MDM2 mediated apoptosis. The rapidity of pAMPK’s activation in response to alcohol, and the ability of interventions targeting AMPK-TORC1-p70/S6K gain- and loss-of-function to alter alcohol’s effects on RBG and p53-mediated apoptosis, suggests that AMPK activation and its downstream suppression of TORC1-p70/S6K are primary effectors of this alcohol-induced apoptosis. These actions of alcohol upon the AMPK-TORC1-p70/S6K pathway to instigate neural crest apoptosis are summarized in Figure 7.
A potential involvement of TORC1 in alcohol-mediated neural crest death has been suggested from prior studies, but those observations have not yet been integrated into the larger, cohesive mechanism investigated here. Specifically, in a zebrafish model of PAE, haploinsufficiency in Pdgfr synergized with alcohol to cause craniofacial deficits, and these were attenuated by treating the embryos with the TORC1 activator L-leucine [46], as TORC1 can operate downstream from PDGFR. Because that study did not directly access TORC1 activity, alcohol’s potential impact on TORC1 in that model is unclear. Acute alcohol exposures reduce TORC1 signaling in adult cerebral cortex [47], similar to its effects here. Using the same antibody reagent as used here (#2971, Cell Signaling), chronic exposure reduced p-mTOR in fetal hippocampus [48]; however, this was accompanied by elevated p-p70/S6K and p4E-BP1, along with elevated Deptor and Rictor but not Raptor, outcomes distinct from those reported here. It is possible that chronic alcohol exposure led to adaptive or compensatory changes in response to reduced p-mTOR or may have dysregulated aspects of that pathway [48]. Our findings are consistent, however, with the increased ULK/ATG signals in alcohol-exposed cells [49,50], as autophagy also increases under TORC1 suppression, and this has been posited to reflect a protective response to alcohol [49,50]; we did not assess autophagy here. Acute alcohol (18-24hr) also suppresses the phosphorylation of the downstream TORC1 targets p70/S6K and 4E-BP1 in SH-SY5Y neuroblastoma cells [51] and C2C12 myocytes [52] and promoted apoptosis in that neuronal model [51]. That TORC1 signaling is indispensable for neural crest is highlighted by studies in which its modulation by genetic [53] or dietary means [54] disrupts craniofacial morphogenesis and alters the facial appearance. Taken together, the critical requirement of neural crest for high rates of RBG to sustain their equally high proliferative rate prior to their migration [25,55,56], and the requirement for TORC1 to sustain p70/S6K activation and RBG [28], could explain how alcohol’s suppression of TORC1 and p70/S6K would enhance their vulnerability to alcohol’s neurotoxicity.
As an anabolic effector of cell growth, TORC1 activity is tightly regulated by both its own direct nutrient sensing and by upstream kinases such as AMPK that further act to integrate metabolic status. As a primary suppressor of TORC1 and RBG [27,43,44], alcohol’s activation of pAMPK and its targets pTSC2 and pRaptor suggest that alcohol reduced RBG via repression of TORC1 signaling. This was further endorsed in that treatment of ALC neural crest with the AMPK antagonist dorsomorphin reversed alcohol’s impact and normalized pAMPK, pS6K, p-mTOR, and pRaptor, and prevented both nucleolar stress and apoptosis. That AMPK inhibition blocked alcohol’s adverse effects indicates that pAMPK participates in their apoptosis. However, that neither rapamycin nor AICAR were sufficient to suppress rRNA synthesis or cause apoptosis in otherwise healthy neural crest suggests that signals in addition to pAMPK and TORC1 contribute to their nucleolar stress and apoptosis. This is not surprising as alcohol does not act through a single receptor, but rather binds and modulates the activity of multiple protein targets [57]. For example, we have previously shown that in cranial neural crest alcohol initiates a G-protein-mediated intracellular calcium transient that activates CaMKII and destabilizes nuclear β-catenin to trigger their apoptosis [15,58–60]. Alcohol also has been shown to modulate facial outcome through signals involving sonic hedgehog, retinoids, and oxidative stress, among others [5,6,11–15]. Studies are underway to elucidate these additional contributors that further destabilize these neural crest cells and tip them into an apoptotic fate.
Alcohol has been shown to increase pAMPK in multiple models including neurons of the hippocampus [61], prefrontal cortex [62], isolated cardiac and myoblast cells [63–65], and hepatocytes [66]. Blockade of AMPK activation prevents neuronal cell death in models of zinc-induced toxicity [67] and Parkinson’s disease [68]. However, how alcohol increases AMPK is currently unclear. AMPK is a master sensor and coordinator of cellular energy status and is activated in response to falling energy status via multiple mechanisms including binding to elevated AMP, carbohydrate binding motifs and, of relevance to this model, pCAMKK2 in response to intracellular calcium signals [43,44,69]. There is debate whether AMPK activation in response to alcohol represents a beneficial or deleterious effect [70], and studies are contradictory in this regard. AMPK can increase glycolytic flux in response to hypoxia, as well as acting to limit energy-demanding processes such as RBG [43,44]. This latter may be especially detrimental to cranial neural crest given their absolute dependence upon RBG as evidenced in the genetic disorders known as ribosomopathies, in which RBG loss-of-function causes p53-mediated neural crest losses and facial deficits that parallel those of PAE (i.e. Treacher-Collins, Diamond-Blackfan Anemia) [34–36]. What is unclear is why alcohol’s suppression of RBG invokes a nucleolar stress instead of a ‘soft landing’ or adaptive response that protects against p53 activation and MDM2/p53-mediated apoptosis, and investigations into this are underway.
In conclusion, pharmacologically relevant alcohol exposures significantly enhance the phosphorylation of AMPK in cranial neural crest and inhibit the downstream TORC1 signaling pathway to reduce p-p70/S6K and p-4E-BP1. This is accompanied by a suppression of de novo rRNA synthesis, induction of nucleolar stress, and stabilization of p53 to initiate their apoptosis. These findings underscore the pivotal role of the AMPK/TORC1 signaling axis as a novel mechanism in mediating alcohol’s deleterious effects on neural crest development and survival.
Funding
This work was supported by the National Institutes of Health [grant number R01 AA011085] and internal funds from the UNC Nutrition Research Institute.
Author Contributions
YH - Conceptualization, methodology, formal analysis, investigation, writing original draft, writing edit and review, visualization; GRF - Conceptualization, methodology, investigation, writing edit and review; SMS - Conceptualization, writing original draft, writing edit and review, visualization, administration, funding.
Acknowledgements
We are grateful to Josh Bausch for early work on this project, and to Thomas Wilkie for helpful discussions.
Footnotes
This version reflects the updated submission that is currently under peer-review.