Skip to main content
NCBI home page
iScience logoLink to iScience
. 2018 Oct 12;9:1–13. doi: 10.1016/j.isci.2018.10.006

AMP-Activated Protein Kinase Is Essential for the Maintenance of Energy Levels during Synaptic Activation

Claudia Marinangeli 1, Sébastien Didier 1, Tariq Ahmed 2,3,4, Raphaelle Caillerez 1, Manon Domise 1, Charlotte Laloux 5, Séverine Bégard 1, Sébastien Carrier 1, Morvane Colin 1, Philippe Marchetti 1, Bart Ghesquière 6, Detlef Balschun 2,3, Luc Buée 1, Jérôme Kluza 1, Valérie Vingtdeux 1,7,∗
PMCID: PMC6203244  PMID: 30368077

Summary

Although the brain accounts for only 2% of the total body mass, it consumes the most energy. Neuronal metabolism is tightly controlled, but it remains poorly understood how neurons meet their energy demands to sustain synaptic transmission. Here we provide evidence that AMP-activated protein kinase (AMPK) is pivotal to sustain neuronal energy levels upon synaptic activation by adapting the rate of glycolysis and mitochondrial respiration. Furthermore, this metabolic plasticity is required for the expression of immediate-early genes, synaptic plasticity, and memory formation. Important in this context, in neurodegenerative disorders such as Alzheimer disease, dysregulation of AMPK impairs the metabolic response to synaptic activation and processes that are central to neuronal plasticity. Altogether, our data provide proof of concept that AMPK is an essential player in the regulation of neuroenergetic metabolic plasticity induced in response to synaptic activation and that its deregulation might lead to cognitive impairments.

Subject Areas: Neuroscience, Molecular Neuroscience, Cellular Neuroscience

Graphical Abstract

graphic file with name fx1.jpg

Open in a new tab

Highlights

  • •

    AMPK is rapidly activated following synaptic activation

  • •

    AMPK stimulates neuronal glycolysis and oxidative respiration, i.e., metabolic plasticity

  • •

    Metabolic plasticity ensures the expression of IEGs and long-term memory formation

  • •

    AMPK deregulation, as in Alzheimer disease, prevents metabolic plasticity response

Neuroscience; Molecular Neuroscience; Cellular Neuroscience

Introduction

Although the brain accounts for only 2% of the total body mass, it is one of the most energy-consuming organs, requiring up to 20% of the oxygen and 25% of the glucose metabolized by the body. These substrates are converted to ATP through glycolysis and oxidative phosphorylation, the latter taking place in mitochondria. Hence, around 20% of the ATP produced in the body is devoted to the brain. Importantly, brain functions depend on an adequate energy supply. Indeed, reducing the brain supply of glucose or oxygen even for short periods leads to severe brain damage, as observed following ischemic stroke. Numerous reports support a role for glucose as a potent modulator of learning and memory, including behavioral tasks, in both humans and rodents. Indeed, both peripheral and direct central glucose administration enhances cognitive processes (Gold, 1995). Similarly, astrocytes-derived lactate, the end product of glycolysis, was reported to be necessary for the formation of long-term memory (Suzuki et al., 2011). Consistent with these observations, brain imaging approaches are based on the detection of signals that are directly related to energy supply and use. For instance, positron emission tomography (PET) reveals changes in cerebral blood flow, glucose uptake, and oxygen consumption according to the tracer that is used. These imaging approaches have been instrumental in demonstrating that brain activity is coupled to increased energy availability and consumption. Further studies have demonstrated that 85% of the energy required by the brain is used by neurons and is devoted to glutamatergic transmission (Attwell and Laughlin, 2001, Harris et al., 2012).

Failure to maintain proper energy levels in the brain also leads to neurodegeneration. The main neurodegenerative diseases are characterized by dysregulations of energy metabolism (Ferreira et al., 2010). For instance, Alzheimer disease (AD) is characterized by mitochondrial dysfunctions (Lin and Beal, 2006) and early glucose hypometabolism (Mosconi, 2005). Of note, [18F] fluorodeoxyglucose (FDG)-PET, a marker for glucose utilization, was recently described as being the most accurate biomarker for AD. Indeed, FDG-PET allowed to correctly differentiate subjects who progressed to AD or other dementia from those who remained stable or reverted to normal cognition (Caminiti et al., 2018). Finally, although these metabolic impairments were proposed to play a causative role in disease pathogenesis, their causes and consequences remain to be established.

One of the most important cell energy sensors and regulators is the AMP-activated protein kinase (AMPK). AMPK is a serine/threonine protein kinase composed of three subunits. It possesses one catalytic subunit alpha and two regulatory subunits beta and gamma. The gamma subunit holds adenine-nucleotide-binding sites that allow the sensing of intracellular levels of AMP, ADP, and ATP. Consequently, any energetic stress that will modify the intracellular AMP/ATP ratio will enhance the activation of AMPK (Xiao et al., 2007). The activation of AMPK requires its phosphorylation on threonine 172 within its catalytic alpha subunit. Two main kinases were reported to fulfill this role, the liver kinase B1, which is thought to be constitutively activated and involved in AMPK phosphorylation in energy stress conditions (Hawley et al., 2003, Woods et al., 2003), and the Ca2+/calmodulin-dependent protein kinase kinase-beta (CaMKKβ), whose activation is dependent on intracellular calcium levels (Hawley et al., 2005). AMPK is highly expressed in the brain, and in particular, in neurons. Interestingly, several studies have reported an over-activation of neuronal AMPK in the brain of patients with AD (Vingtdeux et al., 2011), Parkinson disease (Jiang et al., 2013), and Huntington disease (Ju et al., 2011), thus supporting the defective metabolism hypothesis in these disorders (Domise and Vingtdeux, 2016).

Although it is obvious that brain energy metabolism must be tightly regulated to sustain neuronal function and cognition, the molecular mechanisms involved in this regulation remain poorly understood. In particular, how neurons meet their energy demand to sustain synaptic transmission remains to be established. In this study, we assessed the possibility that AMPK might play a critical role in the regulation of energy metabolism in neurons in response to their activation.

Results

AMPK Is Stimulated following Synaptic Activation

To assess the molecular pathways involved in neuronal energy level maintenance, we used bicuculline and 4-aminopyridine (Bic/4-AP) to induce synaptic release of glutamate in differentiated primary neurons. This well-established protocol (Hardingham et al., 2002, Hoey et al., 2009) is known to preferentially induce glutamatergic neurotransmission that we will refer to hereafter as synaptic activation (SA). To validate the efficiency of this protocol in our model, we first assessed the activation status of a signaling pathway known to be activated following SA, the MAPK pathway. SA induction in 15 days in vitro differentiated mouse primary neurons led to a rapid phosphorylation of the main MAPK pathway components, ERK, RSK, MSK, and MEK, which are representative of this signaling pathway activation (Figure S1A). In addition, blockade of glutamatergic N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors by the specific inhibitors MK-801 and NBQX, respectively, completely abolished the activation of ERK, confirming that MAPK activation is dependent on the stimulation of glutamate receptors (Figure S1B). Together, these results confirm the validity of the SA protocol in our model. To allow a more general screening of the signaling pathways activated following SA we used a kinase array approach. After 5 min Bic/4-AP stimulation, we detected an increase in the phosphorylation status of ERK, MSK, and CREB, as expected from our previous data (Figures 1A, 1B, and 1E). Interestingly, an increase in the phosphorylation levels of AMPK was also observed following Bic/4-AP treatment. AMPK activation was further assessed by western blot using phospho-specific antibodies directed against the AMPK-Thr172 epitope, which is a prerequisite for AMPK activity, and acetyl-CoA carboxylase (ACC)-Ser79, a direct AMPK target (Figures 1G–1I). Following SA, we observed a rapid phosphorylation of AMPK and ACC, thus demonstrating that the AMPK signaling pathway was activated (Figures 1G–1I). This activation was also dependent on glutamate receptor activation since their inhibition significantly reduced AMPK activation (Figures S1B–S1D). Given that SA leads to intracellular calcium influx and that AMPK was reported to be phosphorylated by the calcium-dependent protein kinase CaMKKβ in neuronal cells, we used the CaMKKβ-specific inhibitor STO-609 to address this possibility. However, our results showed that in our model CaMKKβ was not responsible for the SA-induced AMPK activation (Figure S2), thus strongly suggesting that AMPK phosphorylation was not related to intracellular calcium levels. Since SA is a very energetic process, it is likely that AMPK activation might be related to changes in ATP levels. To assess the importance of energy level maintenance for the signaling pathways induced by SA, primary neurons were pre-treated with a combination of oligomycin, an ATP synthase inhibitor, and 2-deoxy-D-glucose, a glycolysis inhibitor (Oligo/2-DG) to deplete intracellular ATP (Figure 1F). In these conditions of energy depletion, SA failed to activate the MAPK signaling pathway (Figures 1C–1E). However, AMPK activity was increased in both conditions (Figures 1C–1E). Altogether, our data show that the activation of the MAPK signaling pathway by SA is contingent upon intracellular energy levels, suggesting a role of AMPK in their regulation.

Figure 1.

Figure 1

Open in a new tab

AMPK Is Activated upon Synaptic Activation

(A–D) 15 days in vitro (DIV) differentiated primary neurons were stimulated with Bic/4-AP (5 min) in the presence or absence of oligomycin and 2-deoxy-D-glucose (Oligo/2-DG, 5-min pre-treatment 1 μM/50 mM). Cell extracts were then probed on phosphoprotein arrays ((A)–(D), upper boxes show phosphorylated ERK1/2 [Thr202/Tyr204, Thr185/Tyr187] and lower boxes show phosphorylated AMPK [Thr172]).

(E) Quantification of the kinase array represented as a heatmap; results are expressed as percentage of the control (Ctrl).

(F) Intracellular ATP quantifications following 5-min Oligo/2-DG treatment (n = 3).

(G) Western blot (WB) analysis of phosphorylated ACC and AMPK and total ACC, AMPK, and actin in lysates obtained from 15 DIV neurons stimulated with Bic/4-AP for the indicated times. WB is representative of at least 4 independent experiments.

(H and I) WB quantifications showing the ratios of phosphorylated AMPK/total AMPK (pAMPK/AMPK) (H) and phosphorylated ACC/total ACC (pACC/ACC) (I) following Bic/4-AP stimulation (n = 4).

Results show means ± SD. Student's t test (F) and one-way ANOVA with Bonferonni post-hoc test (H–I) were used for evaluation of statistical significance. *p < 0.05, **p < 0.01, ***p < 0.001. See also Figures S1 and S2.

AMPK Activation Is Necessary to Induce Neuronal Metabolic Plasticity in Response to Synaptic Activation

To test whether AMPK was involved in the maintenance of energy levels during SA, we assessed metabolic fluxes in differentiated live neurons using the extracellular flux analyzer Seahorse XFe24 (Marinangeli et al., 2018). Glycolytic flux was determined by measuring the extracellular acidification rate (ECAR) due to H+ release in combination with lactate, and mitochondrial respiration was determined by measuring the oxygen consumption rate (OCR). Following Bic/4-AP stimulation, ECAR and OCR were increased, showing that both glycolysis and mitochondrial respiration were enhanced to maintain energy levels following SA (Figures 2A and 2G, green arrows, and Figures 2C and 2I). Several metabolic parameters were determined using calculations described in the Methods section. Interestingly, Bic/4-AP stimulation did not change the overall glycolytic reserve (Figure 2E), maximal respiration (Figure 2K), or spare respiratory capacity (Figure 2L). However, in the presence of Bic/4-AP stimulation, the cells engage a larger fraction of the existing glycolytic reserve (Figure 2F) and spare respiratory capacity (Figure 2M) to increase energy production (Figure 2J). To determine the involvement of AMPK in this metabolic up-regulation induced following SA, neurons were pre-treated with the AMPK inhibitor compound C (Cc). Our results showed that AMPK inhibition prohibited the up-regulation of glycolysis and mitochondrial respiration that were induced by SA (Figures 2A–2M). Although AMPK inhibition in basal conditions (without SA induction) did not affect overall basal glycolysis or mitochondrial functions (Figure S3), it significantly reduced the use of the spare glycolytic reserve and spare respiratory capacity induced by SA (Figures 2F and 2M). Similar results were obtained using a kinase-dead dominant-negative construct of AMPK therefore excluding any potential off-target effect of Cc (Figures S4A and S4B). Finally, the phenogram analysis, which defines the overall cellular metabolic state of the cells, demonstrated that SA induced by Bic/4-AP led to a shift toward a metabolic state of the neurons (Figure 2N) that we will refer to as their metabolic plasticity. AMPK inhibition using Cc almost completely abolished this shift (Figure 2N), thereby suggesting that AMPK activity is necessary to maintain energy levels during SA. To confirm this, intracellular ATP levels were quantified in neurons at different time points following SA in the presence or absence of Cc. In control and SA conditions ATP levels remained constant, whereas AMPK repression led to a significant rapid drop in ATP levels in conditions of SA (Figure 2O). These results showed that AMPK activity following SA is essential to maintain neuronal energy levels.

Figure 2.

Figure 2

Open in a new tab

AMPK Maintains Energy Levels by Up-regulating Glycolysis and Mitochondrial Respiration during Synaptic Activity

(A–M) Extracellular acidification rate (ECAR, A) and oxygen consumption rate (OCR, G) measured using the Seahorse XFe24 technology after Bic/4-AP stimulation (green arrows) in the presence or absence of the AMPK inhibitor compound C (Cc, 10 μM). (A) ECAR profile is monitored under basal conditions and upon the sequential injection of saturating concentration of glucose, followed by Bic/4-AP, or assay medium in control (Ctrl) group, oligomycin (Oligo), and 2-deoxy-D-glucose (2-DG) as indicated by arrows. (G) OCR profile monitored under basal condition and following the sequential injection of assay medium, in Ctrl group, or Bic/4-AP, in the stimulated groups, Oligo, carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), and rotenone and antimycin A (Rot/AA) as indicated by arrows. ECAR and OCR are indicators of glycolysis and mitochondrial respiration, respectively. (B) Basal glycolysis, (C) glycolytic response to treatment, (D) glycolytic capacity, (E) spare glycolytic reserve expressed as mpH/min, (F) spare glycolytic reserve after treatment expressed as percentage of baseline, (H) basal respiration, (I) oxphos response to treatment, (J) ATP turnover coupled to treatment, (K) maximal respiration, (L) spare respiratory capacity expressed as pM O2/min, and (M) spare respiratory capacity after treatment expressed as percentage of baseline were calculated as described in the Methods section; n = 4–5.

(N) Phenogram obtained by plotting the values obtained from the study of glycolysis (A) and mitochondrial respiration (G) following Bic/4-AP stimulation.

(O) Intracellular ATP levels in neurons stimulated with Bic/4-AP for the indicated times in the presence or absence of Cc (n = 3–4). Values are expressed as percentage of ATP in the Ctrl at time 0.

Results show mean ± SD. One-way ANOVA (B–F and H–M) and two-way ANOVA (O) followed by Bonferroni's post-hoc test were used for the evaluation of statistical significance. *p < 0.05, **p < 0.01, ***p < 0.001. See also Figures S3–S5 and S7.

AMPK Up-regulates Mitochondrial Respiration in Response to Synaptic Activation

Glycolysis and mitochondrial respiration are tightly coupled together. Given that neurons are mainly oxidative cells, we asked whether the increase in mitochondrial respiration induced by SA was dependent on glycolysis up-regulation or whether AMPK could have a more direct impact on mitochondria. To assess the latter, we investigated its involvement in the oxidation of mitochondrial substrates that do not require prior glycolysis. As neurons were reported to use lactate as an alternative substrate to glucose in the context of the astrocyte-neuron lactate shuttle, we substituted glucose with lactate in the Seahorse XFe24 experiments. In these conditions, SA up-regulated mitochondrial respiration (Figures S5A–S5F), as it was previously observed with glucose. This metabolic response to SA was inhibited by the presence of Cc, showing that AMPK is also involved in the regulation of lactate utilization following SA. Importantly, these results support a more direct role of AMPK in mitochondrial function. Given that neurons were also reported to use glutamine for energy (Divakaruni et al., 2017), we repeated these experiments with a media devoid of glucose. Results showed that following SA, neurons did not use the glutamine present in the culture media to increase their energy production (Figures S5G–S5L). To go further, we performed metabolomics to determine the fluxes throughout the central carbon metabolism using 13C stable-isotope tracing. For this, neurons were cultured in the presence of uniformly 13C-labeled glucose, and fractional contribution was quantified using mass spectrometry (isotopologs were designated as m0, m1, m2, m3, …). An increase in labeled lactate confirmed the increased glycolytic flow under SA conditions that was prohibited by Cc (Figure 3A). Furthermore, analysis of the isotopolog profiles of the tricarboxylic acid cycle (TCA) metabolites demonstrated significant changes (Figures 3B–3H). Of particular interest is the significant increase in the levels of m3 isotopomers in SA from citrate, succinate, fumarate, and malate, suggesting that SA induced a shift toward pyruvate carboxylation (Buescher et al., 2015). Higher abundances of m4, m5, and m6 isotopologs from citrate were observed following Bic/4-AP stimulation (Figure 3C). These isotopomers require the combined action of pyruvate dehydrogenase (PDH), pyruvate carboxylase (PC), and multiple turns of the Kerbs cycle, as explained in Figure 3B. These results suggest that Bic/4-AP stimulation induced a shift toward anaplerotic pyruvate carboxylation that was not observed when AMPK activity was repressed. Altogether, these data support the notion that AMPK activity is needed to induce the metabolic “reprogramming” that occurs following SA.

Figure 3.

Figure 3

Open in a new tab

AMPK Up-regulates Mitochondrial Metabolism upon Synaptic Activation

Measurement of 13C-glucose-derived metabolites, calculated as percentage of the total metabolite pool following liquid chromatography-mass spectrometry analysis of primary neurons after 30 min in culture with 13C-glucose-supplemented media with or without Bic/4-AP stimulation.

(A) Metabolic profile of glycolysis intermediates.

(B) Schematic representation of the labeling patterns of metabolic intermediates starting from uniformly labeled glucose. The letter “m” indicates the number of carbon atoms labeled with 13C. The blue filled circles represent 13C atoms derived from pyruvate dehydrogenase activity, whereas the dashed circles represent 13C atoms derived from the pyruvate carboxylase activity.

(C–H) Isotopolog quantification of the Krebs cycle intermediates citrate (C), isocitrate (D), α-ketoglutarate (E), succinate (F), fumarate (G), and malate (H). Metabolic profiles are expressed as heatmaps. n = 3.

See also Figure S5.

Neuronal Metabolic Plasticity Induced by AMPK Is Crucial for Immediate-Early Genes Expression, Synaptic Plasticity, and Long-Term Memory Formation

We next sought to determine the importance of this metabolic plasticity regulated by AMPK for neuronal functions. SA, when sustained, leads to the expression of immediate-early genes (IEGs) that include Arc, cFos, and Egrl. These genes are known to modulate the neuronal plasticity underlying learning and memory. In our model of SA, we observed the expression of these IEGs starting 1 hr after Bic/4-AP treatment (Figure 4A). This induction of IEG expression was inhibited when the energetic status was disrupted by either the inhibition of glycolysis with the glyceraldehyde-3-phosphate dehydrogenase inhibitor iodoacetate or 2-DG or the interruption of mitochondrial respiration using oligomycin and the inhibitors of complexes I and III rotenone and antimycin A, respectively (Figure 4B). These results demonstrated that the maintenance of energy levels is crucial for the expression of IEGs. In addition, AMPK inhibition also led to a significant reduction of IEG expression following SA (Figures 4C and S4C), showing that AMPK-mediated metabolic plasticity is necessary for the expression of IEGs following SA.

Figure 4.

Figure 4

Open in a new tab

AMPK Activation Following Synaptic Activity Is Necessary for IEG Expression, Synaptic Plasticity, and Long-Term Memory Formation

(A) Western blot (WB) analysis of the IEGs Arc, EgrI, and cFos expression following Bic/4-AP stimulation for the indicated times in differentiated primary neurons at 15 DIV (days in vitro). WB is representative of at least 5 independent experiments.

(B) WB analysis of the IEGs Arc, EgrI, and cFos expression following 2 hr Bic/4-AP stimulation in the presence or absence of glycolysis inhibitors (2-DG 50 mM, Iodo 100 μM pre-treatment of 5 min) or mitochondrial respiration inhibitors (Oligo 1 μM, rotenone [Rot] 1 μM, antimycin A [AA] 1 μM pre-treatment of 5 min) in differentiated primary neurons at 15 DIV. WB is representative of at least 4 independent experiments.

(C) WB analysis of the IEGs Arc, EgrI, and cFos expression following 2 hr Bic/4-AP stimulation in the presence or absence of Cc (20 min pre-treatment, 10 μM). WB is representative of at least 4 independent experiments.

(D) Recording of LTP in the CA1 region of the hippocampus ex vivo. Triple application of an electrical theta burst stimulation protocol induced robust LTP in controls. However, in slices treated with 10 μM Cc, the same stimulation generated only an impaired, decremental potentiation, returning to baseline values after about 80 min (Wilcoxon test). This resulted in a highly significant difference (F1,13 = 120.105; p < 0.001, RM-ANOVA). Data are presented as mean ± SEM, where n refers to the number of animals tested.

(E) Schematic representation of the IA protocol timeline.

(F) Hippocampal injections of Cc 20 min before the IA training (Acq) disrupted long-term memory at 24 hr (Test 1). The disruption persisted 6 days after training (Test 2). Cc-injected mice had normal retention after retraining (Test 3). n = 10, results show mean ± SEM.

One-way ANOVA followed by Bonferroni's post-hoc test was used for evaluation of statistical significance. ***p < 0.001. See also Figure S4.

To examine the impact of AMPK inhibition on synaptic plasticity under more physiological conditions, we took advantage of ex vivo hippocampal slices. Thus, we investigated a putative role of AMPK in long-term potentiation (LTP) in the CA1 region of the hippocampus, a well-established model of learning and memory at the cellular level. As depicted in Figure 4D, triple application of an electrical theta burst stimulation protocol resulted in robust LTP in control (vehicle) with initial values of 213.3 ± 5.3% (n = 7), which were retained at about this level until 4 hr after induction (197.5 ± 15.1%). This robust type of LTP is protein synthesis dependent and therefore a relevant model for the formation of long-term memory (Ahmed and Balschun, unpublished data). Application of 10 μM Cc, in contrast, caused a severe impairment of LTP with a significantly lower initial magnitude of 146.2 ± 9.4% (p = 0.0001 Welch test) and a decremental potentiation that returned to baseline values after about 80 min (Wilcoxon test). Overall, this resulted in a highly significant difference in LTP in control mice (F1,13 = 120.105; p < 0.001, repeated measures (RM)-ANOVA) underlining an essential function of AMPK in long-term synaptic plasticity. Furthermore, quantification of the LTP defect, in percentage, of control for the average of first 30 min used as a measure of induction (vehicle LTP: 200% ± 7%, n = 11; Cc LTP: 126% ± 6%, n = 8; RM-ANOVA [F1,17] = 25.58, p < 0.0001) and for the average of 210–240 min used as a measure of late maintenance (vehicle LTP: 159% ± 9%, n = 11; Cc LTP: 77% ± 11%, n = 8; RM-ANOVA [F1,17] = 25.58 p < 0.0001) demonstrated that Cc reduced both the induction and the late maintenance phases.

Finally, to evaluate the importance of AMPK-mediated metabolic plasticity in vivo, we tested whether AMPK activation is involved in long-term memory retention. To this end, Cc was bilaterally injected in the hippocampus of awake wild-type mice 20 min before inhibitory avoidance (IA) training, a task that depends on the proper functioning of the hippocampus (Izquierdo et al., 2016, Whitlock et al., 2006) (Figures 4E and 4F). Cc injection did not affect IA acquisition, as the mean latencies to enter the shock compartment during training (Acq) were similar in both groups. However, Cc significantly blocked long-term memory tested at 24 hr (Test 1). Retesting 6 days after acquisition (Test 2) showed that the memory loss persisted. Retraining of the Cc injected group resulted in normal memory retention 24 hr later (Test 3), indicating that the hippocampus was functionally intact. Altogether, these results show that AMPK is involved in long-term memory formation and provided proof of concept that AMPK plays an essential role in cognitive processes.

AMPK Deregulation Impairs the Neuronal Metabolic Plasticity Induced by Synaptic Activation

Last, we asked whether AMPK deregulation as it is observed in AD could impair AMPK-mediated metabolic plasticity induced by SA. Since AMPK is over-activated in neurons of patients with AD when compared with control patients (Vingtdeux et al., 2011), we mimicked this condition by treating primary neurons with AICAR, an AMPK agonist, 24 hr before inducing SA (referred to as pre-activation hereafter). AMPK pre-activation with AICAR did not modify the metabolic profile of neurons, whereas it greatly impaired the ability of neurons to up-regulate mitochondrial respiration in response to SA (Figures 5A and 5B). Similar results were obtained after over-expression of a constitutively active mutant of AMPK (CA-AMPK) in neurons (Figure 5C), further supporting the inhibitory effect of AMPK pre-activation on the metabolic response to SA. We then assessed the effect of AMPK pre-activation on neuronal ATP content during SA (Figure S6). Our results showed that, when compared with unstimulated cells, ATP content, both in control and AICAR pre-treated groups remained constant during SA. These results suggest that AMPK pre-activation might have produced metabolic alterations that prevented SA from occurring in a different way than the acute inhibition observed with Cc. In this context, the pre-activation of AMPK either using AICAR or CA-AMPK also prevented the SA-induced expression of IEGs (Figures 5D and 5E). Altogether, our results show that a tight regulation of AMPK is important for the physiological metabolic response of neurons to SA and for the triggering of processes leading to neuronal plasticity.

Figure 5.

Figure 5

Open in a new tab

AMPK Pre-activation Prevents Neuronal Metabolic Response to SA

(A) Mitochondrial respiration profile in primary neurons evaluated using the Seahorse XFe24 technology to investigate the impact of long-lasting (24 hr) AMPK activation induced by AICAR in response to Bic/4-AP. Mitochondrial respiration profile, expressed as oxygen consumption rate (OCR), is monitored upon the injection of assay medium, in control group (Ctrl) and in cells treated with AICAR (AICAR), or upon the injection of Bic/4-AP in the absence (Bic/4-AP) or presence of AICAR (Bic/4-AP + AICAR), followed by injection of Oligo, carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), and rotenone and antimycin A [Rot/AA] as indicated by the arrows. Results are representative of 3 independent experiments.

(B) OCR response to Bic/4-AP stimulation expressed as pMO2/min (n = 3).

(C) OCR response to Bic/4-AP stimulation expressed as pMO2/min in primary neurons over-expressing a constitutively active (CA) form of AMPK (n = 3).

(D) Immunoblotting with phosphorylated ACC (pACC), phosphorylated AMPK (pAMPK), Arc, EgrI, cFos, and actin in lysates obtained from 15 days in vitro (DIV) neurons stimulated with Bic/4-AP for 2 hr after pre-treatment with AICAR (24 hr, 1 mM). Results are representative of at least 5 independent experiments. Quantification of western blot (WB) showing the expression of the IEGs Arc, EgrI, and cFos (n = 5).

(E) Immunoblotting with total AMPK (# CA labels the over-expressed CA-AMPK construct), Arc, EgrI, cFos, and actin in lysates obtained from 15 DIV neurons stimulated with Bic/4-AP for 2 hr, 4 days after injection with CA-AMPK. Results are representative of at least 3 experiments. Quantification of WB showing the expression of the IEGs Arc, EgrI, and cFos (n = 3).

Results show mean ± SD. One-way ANOVA followed by Bonferroni's post-hoc test was used for evaluation of statistical significance. *p < 0.05, ***p < 0.001. See also Figures S6 and S7.

Discussion

Neurons have high energy demands that are met through the oxidation of glucose via glycolysis and mitochondrial respiration. Neuronal activity is a process known to require and thus stimulate ATP production (Rangaraju et al., 2014); however, the molecular mechanisms involved in this regulation remain poorly understood. Here, we demonstrated that the AMPK-mediated signaling pathway plays a key role in this regulation. We found that AMPK is rapidly activated following SA to maintain energy levels via the up-regulation of glycolytic and mitochondrial use of glucose. Regulation of glucose uptake by AMPK in neurons has been previously described. AMPK was proposed to regulate glucose transporters GLUT3 and GLUT4 surface expression following glutamate excitation and action potential firing, respectively (Ashrafi et al., 2017, Weisova et al., 2009). It is possible that a mobilization of glucose transporters to the plasma membrane also occurs in our conditions, which could explain how AMPK regulates glycolysis. Our results, in addition to validate the increased glycolysis following SA, also demonstrate that mitochondrial respiration plays an important part in sustaining energy levels. Indeed, our data demonstrate that AMPK also acts on mitochondrial respiration independent of glycolysis. Actually, consistent with their high energy requirements, neurons sustain a high rate of oxidative metabolism (Boumezbeur et al., 2010). Bearing this in mind, the importance of mitochondria for cognitive functions has recently regained importance. In particular, a recent study has shown that cannabinoid receptors present on mitochondria (mtCB1) are responsible for amnesia induced by acute cannabinoid intoxication. In particular, activation of hippocampal mtCB1 receptors led to memory impairments due to decreased mitochondrial respiration (Hebert-Chatelain et al., 2016). Our study reinforces this link between mitochondrial activity and memory formation. Our results also raised the question of AMPK subcellular localization. Indeed, it could be speculated that a pool of AMPK might be present at or within the mitochondria to up-regulate mitochondrial respiration. Previous reports have also highlighted this possibility (Liang et al., 2015). In our context, further work will be required to determine whether PC or PDH could be direct AMPK substrates. In particular, PC catalyzes the conversion of pyruvate to oxaloacetate, named pyruvate carboxylation, within the mitochondria. Although this step, which leads to de novo synthesis of TCA cycle intermediates is crucial to replenish TCA cycle and supports the formation of transmitter glutamate, is widely described to occur in astrocytes, PC activity was also reported in neurons (for review see [Hassel and Brathe, 2000]). Another way to replenish TCA is via glutaminolysis. Although this possibility cannot be excluded, it seems not to be involved in our model given that the experiments performed without glucose did not show any response to SA, although glutamine was present in cell culture media. Another interesting possibility is the regulation by AMPK of the AKAP1 (A kinase anchor protein 1) protein, which was identified in muscle by a global phosphoproteomic approach (Hoffman et al., 2015). This direct phosphorylation of AKAP1 by AMPK was reported to regulate mitochondrial respiration (Hoffman et al., 2015). Although we assessed only the immediate metabolic response of neurons to synaptic activity, it will also be interesting to determine whether AMPK could be involved in gene expression changes that promote the neuronal Warburg effect in response to prolonged Bic/4-AP stimulation (Bas-Orth et al., 2017).

Our results are also particularly important in the context of neurodegenerative disorders in which energy metabolism is strongly impaired and AMPK is over-activated (Jiang et al., 2013, Ju et al., 2011, Vingtdeux et al., 2011). In conditions of over-activation, AMPK is unlikely to respond properly to SA, perhaps as a way to preserve energy levels. As a consequence, we propose that AMPK might act as a metabolic switch that determines whether SA may or may not occur according to the currently available energy levels (and hence metabolic substrate) in its vicinity. It is also possible that AMPK over-activation could indirectly impair the metabolic plasticity response to SA. Under these conditions, the identification of AMPK targets as well as the consequences of AMPK over-activation on synaptic integrity will be of interest. Supporting this hypothesis, previous reports have determined that AMPK activation by pharmacological drugs led to impairments in synaptic plasticity (Potter et al., 2010) and memory formation (Dash et al., 2006). In addition, in the context of AD, AMPK was proposed to be a key player in the reduction of number of dendritic spine (Mairet-Coello et al., 2013) and in the LTP impairments induced by Aβ oligomers (Ma et al., 2014). These reports show that up-regulation of AMPK activity has the same negative outcome on synaptic plasticity and memory formation as AMPK inhibition.

Altogether, our data suggest that AMPK activity has to be fine-tuned extremely well since any hypo- or hyperactivation will have detrimental consequences on synaptic plasticity and hence memory formation.

In summary, our results demonstrate that AMPK plays a critical role during glutamatergic SA. In particular, we show that AMPK activity is crucial to maintain neuronal energy levels and hence is involved in memory formation processes. Our study establishes AMPK as an important link between memory formation and energy metabolism and suggests that it could act as a metabolic checkpoint by sensing the availability of energy substrates. These findings might also be of value in the context of diabetes and obesity, conditions that are associated with cognitive (Dye et al., 2017, McCrimmon et al., 2012) and central metabolism (Hwang et al., 2017) impairments in humans.

Limitations of Study

Our data also demonstrate that the energetic metabolic up-regulation mediated by AMPK is crucial for the expression of IEGs, synaptic plasticity, and long-term memory formation. In line with these results, AMPK was reported to be activated and necessary for LTP induced in vivo by high-frequency stimulation (Yu et al., 2016). However, our in vivo data were obtained using a pharmacological approach to inhibit AMPK; therefore for a more thorough understanding of these processes, the generation of conditional AMPK knockout mouse models will be necessary to disentangle the importance of neuronal AMPK in particular types and phases of memory.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Acknowledgments

We thank the animal core facility (animal facilities of Université de Lille-Inserm) of “Plateformes en Biologie Santé de Lille” as well as C. Degraeve, M. Besegher-Dumoulin, J. Devassine, R. Dehaynin, and D. Taillieu for animal care. The authors thank the behavioral exploration platform for rodent (Federation of Neurosciences, Univ. Lille, France). pAMPK alpha2 WT and pAMPK alpha2 K45R were a gift from Morris Birnbaum (Addgene plasmids # 15991 and 15992).

This work was supported by the French Fondation pour la cooperation Scientifique—Plan Alzheimer 2008-2012 (Senior Innovative Grant 2013) to V.V., by the Fondation Vaincre Alzheimer (n°FR-16071p) to V.V., and in part by the Labex DISTALZ (Development of Innovative Strategies for a Transdisciplinary Approach to Alzheimer's disease) to L.B.; by INSERM, UNIVERSITE DE LILLE II, Ligue Contre le Cancer (Comité du Nord, Comité de l’Aisne) to P.M. and J.K.; and a special financial support from the Association pour l’Etude des Anomalies Congénitales Neurodev of Pr. B. Poupard to P.M. T.A. and D.B. are supported by the Fonds voor Wetenschappelijk Onderzoek (FWO) Flanderen, project G0D7614N. M.D. holds a doctoral scholarship from Lille 2 University.

Author Contributions

Conceptualization, J.K. and V.V.; Methodology, C.M., T.A., C.L., B.G., D.B., J.K., and V.V.; Investigation, C.M., S.D., T.A., R.C., M.D., B.G., J.K., and V.V.; Resources, S.B., S.C., M.C., P.M., and L.B.; Writing – Original Draft, C.M. and V.V.; Writing – Review & Editing, T.A., P.M., B.G., D.B., L.B., J.K., and V.V.; Supervision, J.K. and V.V.

Declaration of Interests

The authors declare that they have no conflict of interest.

Published: November 30, 2018

Footnotes

Supplemental Information includes Transparent Methods and seven figures and can be found with this article online at https://doi.org/10.1016/j.isci.2018.10.006.

Supplemental Information

Document S1. Transparent Methods and Figures S1–S7
mmc1.pdf (1.2MB, pdf)

References

  1. Ashrafi G., Wu Z., Farrell R.J., Ryan T.A. GLUT4 mobilization supports energetic demands of active synapses. Neuron. 2017;93:606–615.e3. doi: 10.1016/j.neuron.2016.12.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Attwell D., Laughlin S.B. An energy budget for signaling in the grey matter of the brain. J. Cereb. Blood Flow Metab. 2001;21:1133–1145. doi: 10.1097/00004647-200110000-00001. [DOI] [PubMed] [Google Scholar]
  3. Bas-Orth C., Tan Y.W., Lau D., Bading H. Synaptic activity drives a genomic program that promotes a neuronal Warburg effect. J. Biol. Chem. 2017;292:5183–5194. doi: 10.1074/jbc.M116.761106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Boumezbeur F., Mason G.F., de Graaf R.A., Behar K.L., Cline G.W., Shulman G.I., Rothman D.L., Petersen K.F. Altered brain mitochondrial metabolism in healthy aging as assessed by in vivo magnetic resonance spectroscopy. J. Cereb. Blood Flow Metab. 2010;30:211–221. doi: 10.1038/jcbfm.2009.197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Buescher J.M., Antoniewicz M.R., Boros L.G., Burgess S.C., Brunengraber H., Clish C.B., DeBerardinis R.J., Feron O., Frezza C., Ghesquiere B. A roadmap for interpreting (13)C metabolite labeling patterns from cells. Curr. Opin. Biotechnol. 2015;34:189–201. doi: 10.1016/j.copbio.2015.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Caminiti S.P., Ballarini T., Sala A., Cerami C., Presotto L., Santangelo R., Fallanca F., Vanoli E.G., Gianolli L., Iannaccone S. FDG-PET and CSF biomarker accuracy in prediction of conversion to different dementias in a large multicentre MCI cohort. Neuroimage Clin. 2018;18:167–177. doi: 10.1016/j.nicl.2018.01.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Dash P.K., Orsi S.A., Moore A.N. Spatial memory formation and memory-enhancing effect of glucose involves activation of the tuberous sclerosis complex-mammalian target of rapamycin pathway. J. Neurosci. 2006;26:8048–8056. doi: 10.1523/JNEUROSCI.0671-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Divakaruni A.S., Wallace M., Buren C., Martyniuk K., Andreyev A.Y., Li E., Fields J.A., Cordes T., Reynolds I.J., Bloodgood B.L. Inhibition of the mitochondrial pyruvate carrier protects from excitotoxic neuronal death. J. Cell Biol. 2017;216:1091–1105. doi: 10.1083/jcb.201612067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Domise M., Vingtdeux V. AMPK in neurodegenerative diseases. EXS. 2016;107:153–177. doi: 10.1007/978-3-319-43589-3_7. [DOI] [PubMed] [Google Scholar]
  10. Dye L., Boyle N.B., Champ C., Lawton C. The relationship between obesity and cognitive health and decline. Proc. Nutr. Soc. 2017;76:443–454. doi: 10.1017/S0029665117002014. [DOI] [PubMed] [Google Scholar]
  11. Ferreira I.L., Resende R., Ferreiro E., Rego A.C., Pereira C.F. Multiple defects in energy metabolism in Alzheimer's disease. Curr. Drug Targets. 2010;11:1193–1206. doi: 10.2174/1389450111007011193. [DOI] [PubMed] [Google Scholar]
  12. Gold P.E. Role of glucose in regulating the brain and cognition. Am. J. Clin. Nutr. 1995;61:987S–995S. doi: 10.1093/ajcn/61.4.987S. [DOI] [PubMed] [Google Scholar]
  13. Hardingham G.E., Fukunaga Y., Bading H. Extrasynaptic NMDARs oppose synaptic NMDARs by triggering CREB shut-off and cell death pathways. Nat. Neurosci. 2002;5:405–414. doi: 10.1038/nn835. [DOI] [PubMed] [Google Scholar]
  14. Harris J.J., Jolivet R., Attwell D. Synaptic energy use and supply. Neuron. 2012;75:762–777. doi: 10.1016/j.neuron.2012.08.019. [DOI] [PubMed] [Google Scholar]
  15. Hassel B., Brathe A. Neuronal pyruvate carboxylation supports formation of transmitter glutamate. J. Neurosci. 2000;20:1342–1347. doi: 10.1523/JNEUROSCI.20-04-01342.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hawley S.A., Boudeau J., Reid J.L., Mustard K.J., Udd L., Makela T.P., Alessi D.R., Hardie D.G. Complexes between the LKB1 tumor suppressor, STRAD alpha/beta and MO25 alpha/beta are upstream kinases in the AMP-activated protein kinase cascade. J. Biol. 2003;2:28. doi: 10.1186/1475-4924-2-28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hawley S.A., Pan D.A., Mustard K.J., Ross L., Bain J., Edelman A.M., Frenguelli B.G., Hardie D.G. Calmodulin-dependent protein kinase kinase-beta is an alternative upstream kinase for AMP-activated protein kinase. Cell Metab. 2005;2:9–19. doi: 10.1016/j.cmet.2005.05.009. [DOI] [PubMed] [Google Scholar]
  18. Hebert-Chatelain E., Desprez T., Serrat R., Bellocchio L., Soria-Gomez E., Busquets-Garcia A., Pagano Zottola A.C., Delamarre A., Cannich A., Vincent P. A cannabinoid link between mitochondria and memory. Nature. 2016;539:555–559. doi: 10.1038/nature20127. [DOI] [PubMed] [Google Scholar]
  19. Hoey S.E., Williams R.J., Perkinton M.S. Synaptic NMDA receptor activation stimulates alpha-secretase amyloid precursor protein processing and inhibits amyloid-beta production. J. Neurosci. 2009;29:4442–4460. doi: 10.1523/JNEUROSCI.6017-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hoffman N.J., Parker B.L., Chaudhuri R., Fisher-Wellman K.H., Kleinert M., Humphrey S.J., Yang P., Holliday M., Trefely S., Fazakerley D.J. Global phosphoproteomic analysis of human skeletal muscle reveals a network of exercise-regulated kinases and AMPK substrates. Cell Metab. 2015;22:922–935. doi: 10.1016/j.cmet.2015.09.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hwang J.J., Jiang L., Hamza M., Sanchez Rangel E., Dai F., Belfort-DeAguiar R., Parikh L., Koo B.B., Rothman D.L., Mason G. Blunted rise in brain glucose levels during hyperglycemia in adults with obesity and T2DM. JCI Insight. 2017;2:e95913. doi: 10.1172/jci.insight.95913. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Izquierdo I., Furini C.R., Myskiw J.C. Fear memory. Physiol. Rev. 2016;96:695–750. doi: 10.1152/physrev.00018.2015. [DOI] [PubMed] [Google Scholar]
  23. Jiang P., Gan M., Ebrahim A.S., Castanedes-Casey M., Dickson D.W., Yen S.H. Adenosine monophosphate-activated protein kinase overactivation leads to accumulation of alpha-synuclein oligomers and decrease of neurites. Neurobiol. Aging. 2013;34:1504–1515. doi: 10.1016/j.neurobiolaging.2012.11.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ju T.C., Chen H.M., Lin J.T., Chang C.P., Chang W.C., Kang J.J., Sun C.P., Tao M.H., Tu P.H., Chang C. Nuclear translocation of AMPK-alpha1 potentiates striatal neurodegeneration in Huntington's disease. J. Cell. Biol. 2011;194:209–227. doi: 10.1083/jcb.201105010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Liang J., Xu Z.X., Ding Z., Lu Y., Yu Q., Werle K.D., Zhou G., Park Y.Y., Peng G., Gambello M.J. Myristoylation confers noncanonical AMPK functions in autophagy selectivity and mitochondrial surveillance. Nat. Commun. 2015;6:7926. doi: 10.1038/ncomms8926. [DOI] [PubMed] [Google Scholar]
  26. Lin M.T., Beal M.F. Mitochondrial dysfunction and oxidative stress in neurodegenerative diseases. Nature. 2006;443:787–795. doi: 10.1038/nature05292. [DOI] [PubMed] [Google Scholar]
  27. Ma T., Chen Y., Vingtdeux V., Zhao H., Viollet B., Marambaud P., Klann E. Inhibition of AMP-activated protein kinase signaling alleviates impairments in hippocampal synaptic plasticity induced by amyloid beta. J. Neurosci. 2014;34:12230–12238. doi: 10.1523/JNEUROSCI.1694-14.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mairet-Coello G., Courchet J., Pieraut S., Courchet V., Maximov A., Polleux F. The CAMKK2-AMPK kinase pathway mediates the synaptotoxic effects of Abeta oligomers through Tau phosphorylation. Neuron. 2013;78:94–108. doi: 10.1016/j.neuron.2013.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Marinangeli C., Kluza J., Marchetti P., Buee L., Vingtdeux V. Study of AMPK-regulated metabolic fluxes in neurons using the seahorse XFe analyzer. Methods Mol. Biol. 2018;1732:289–305. doi: 10.1007/978-1-4939-7598-3_19. [DOI] [PubMed] [Google Scholar]
  30. McCrimmon R.J., Ryan C.M., Frier B.M. Diabetes and cognitive dysfunction. Lancet. 2012;379:2291–2299. doi: 10.1016/S0140-6736(12)60360-2. [DOI] [PubMed] [Google Scholar]
  31. Mosconi L. Brain glucose metabolism in the early and specific diagnosis of Alzheimer's disease. FDG-PET studies in MCI and AD. Eur. J. Nucl. Med. Mol. Imaging. 2005;32:486–510. doi: 10.1007/s00259-005-1762-7. [DOI] [PubMed] [Google Scholar]
  32. Potter W.B., O'Riordan K.J., Barnett D., Osting S.M., Wagoner M., Burger C., Roopra A. Metabolic regulation of neuronal plasticity by the energy sensor AMPK. PLoS One. 2010;5:e8996. doi: 10.1371/journal.pone.0008996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rangaraju V., Calloway N., Ryan T.A. Activity-driven local ATP synthesis is required for synaptic function. Cell. 2014;156:825–835. doi: 10.1016/j.cell.2013.12.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Suzuki A., Stern S.A., Bozdagi O., Huntley G.W., Walker R.H., Magistretti P.J., Alberini C.M. Astrocyte-neuron lactate transport is required for long-term memory formation. Cell. 2011;144:810–823. doi: 10.1016/j.cell.2011.02.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Vingtdeux V., Davies P., Dickson D.W., Marambaud P. AMPK is abnormally activated in tangle- and pre-tangle-bearing neurons in Alzheimer's disease and other tauopathies. Acta Neuropathol. 2011;121:337–349. doi: 10.1007/s00401-010-0759-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Weisova P., Concannon C.G., Devocelle M., Prehn J.H., Ward M.W. Regulation of glucose transporter 3 surface expression by the AMP-activated protein kinase mediates tolerance to glutamate excitation in neurons. J. Neurosci. 2009;29:2997–3008. doi: 10.1523/JNEUROSCI.0354-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Whitlock J.R., Heynen A.J., Shuler M.G., Bear M.F. Learning induces long-term potentiation in the hippocampus. Science. 2006;313:1093–1097. doi: 10.1126/science.1128134. [DOI] [PubMed] [Google Scholar]
  38. Woods A., Johnstone S.R., Dickerson K., Leiper F.C., Fryer L.G., Neumann D., Schlattner U., Wallimann T., Carlson M., Carling D. LKB1 is the upstream kinase in the AMP-activated protein kinase cascade. Curr. Biol. 2003;13:2004–2008. doi: 10.1016/j.cub.2003.10.031. [DOI] [PubMed] [Google Scholar]
  39. Xiao B., Heath R., Saiu P., Leiper F.C., Leone P., Jing C., Walker P.A., Haire L., Eccleston J.F., Davis C.T. Structural basis for AMP binding to mammalian AMP-activated protein kinase. Nature. 2007;449:496–500. doi: 10.1038/nature06161. [DOI] [PubMed] [Google Scholar]
  40. Yu D.F., Shen Z.C., Wu P.F., Guan X.L., Chen T., Jin Y., Hu Z.L., Ni L., Wang F., Chen J.G. HFS-triggered AMPK activation phosphorylates GSK3β and induces E-LTP in rat hippocampus in vivo. CNS Neurosci. Ther. 2016;22:525–531. doi: 10.1111/cns.12532. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Transparent Methods and Figures S1–S7
mmc1.pdf (1.2MB, pdf)

Articles from iScience are provided here courtesy of Elsevier

RESOURCES