This study investigated the metabolic requirements for neuronal progenitor maintenance in vitro and in vivo by examining the metabolic adaptations that support neuronal progenitors and neural stem cells (NSCs) in their undifferentiated state. We demonstrate that neuronal progenitors are strictly dependent on lactate metabolism, while glucose induces their neuronal differentiation. Lactate signaling is not by itself capable of maintaining the progenitor phenotype. The consequences of lactate metabolism include increased mitochondrial and oxidative metabolism, with a strict reliance on cataplerosis through the mitochondrial phosphoenolpyruvate carboxykinase (PEPCK-M) pathway to support anabolic functions, such as the production of extracellular matrix. In vivo, lactate maintains/induces populations of postnatal neuronal progenitors/NSCs in a PEPCK-M-dependent manner. Taken together, our data demonstrate that, lactate alone or together with other physical/biochemical cues maintain NSCs/progenitors with a metabolic signature that is classically found in tissues with high anabolic capacity.
Glucose is still considered the major fuel for the adult brain, even though lactate signaling or utilization by neurons has been commonly observed both in vitro and in vivo (Bouzier-Sore et al. 2003; Gallagher et al. 2009; Wohnsland et al. 2010). This is especially the case in the developing brain, where lactate is a major substrate for oxidative metabolism in addition to being selectively utilized as an anabolic source for cell proliferation and differentiation (Bolanos and Medina 1993; Nehlig and Pereira de Vasconcelos 1993; Medina et al. 1996; Medina and Tabernero 2005; Zilberter et al. 2010). In neurons, lactate is transported across the cell membrane by the monocarboxylate transporter 2 (MCT2) (Pellerin et al. 1998; Debernardi et al. 2003) and subsequently converted to pyruvate, which can be further oxidized or used as an initial substrate in anabolic pathways. In contrast, astroglial cells release lactate to the extracellular milieu through the transporters MCT1 and MCT4 (Rafiki et al. 2003; Pierre and Pellerin 2005). Lactate is also involved in the maintenance of brain energy turnover and neurovascular coupling, by regulating cyclic AMP formation through GPR81 receptor activation, a process independent of lactate metabolism (Bergersen and Gjedde 2012; Bozzo et al. 2013; Lauritzen et al. 2013).
Tissue remodeling and regeneration depend on adult stem cells and their progenitor properties, which are mostly regulated by the cellular microenvironment or niche (Scadden 2006; Ivanovic 2009; Nakada et al. 2011; Gattazzo et al. 2014). Neural stem cells (NSCs) of the central nervous system reside in niches that are in close contact with the vasculature (Shen et al. 2008; Tavazoie et al. 2008; Goldman and Chen 2011). These neurogenic neurovascular niches contain a extracellular matrix (ECM) rich in laminins, type IV collagen, nidogen, and highly glycosylated sulfate proteoglycans (Yanagisawa and Yu 2007; Miner 2008; Shen et al. 2008; Nasu et al. 2012). The balance between cellular self-renewal and differentiation is mediated by cell adhesion to the ECM, and by local cytokines, systemic hormones, and oxygen and nutrient supplies (De Filippis and Delia 2011; Lehtinen et al. 2011; Nakada et al. 2011; Ziegler et al. 2012).
Our previous work showed that neuronal cultures grown on lactate-releasing biomimetic materials composed of poly-l/dl-lactate or treated with l-lactate were maintained in a less differentiated state in which a significant pool of neuronal progenitors was preserved (Alvarez et al. 2013). Our group has also shown that the biophysical properties and topography of the supplied substrate are important for NSC and glial progenitor maintenance (Mattotti et al. 2012; Alvarez et al. 2014). In this study, we explore the role of lactate in the NSC niche by examining the differential effects of lactate and glucose on cellular fate, metabolism, and anabolic capacities of neural progenitors in vitro and in vivo. Our results provide evidence that lactate intake through MCT2 and its subsequent oxidative metabolism direct progenitor commitment to a neuronal progenitor fate. The consequences of lactate metabolism are increased mitochondrial oxidation and a dependence on cataplerosis through the mitochondrial phosphoenolpyruvate carboxykinase (PEPCK-M) pathway. PEPCK-M enhances tricarboxylic acid (TCA) cycle flux and phosphoenolpyruvate (PEP) export to feed-forward carbons into the cytosolic 3-carbon pool, thus effectively promoting the cellular anabolic potential and ECM production. Taken together, these results indicate the coexistence in the neurovascular niche of NSC and progenitors with different metabolic signatures.
Materials and Methods
Intraventricular Brain Injections in Postnatal Animals
All animal housing and procedures were approved by the Institutional Animal Care and Use Committee of our institution in accordance with Spanish and EU regulations.
The brains of newborn (postnatal day 0, P0) mice were injected in the lateral ventricle with 2 µL of lactic acid (5 mM, Sigma-Aldrich, St Louis, MO, n = 10), MCT1/2 inhibitor AR-C155858 (100 nM, Adooq, n = 10), 3-mercaptopicolinic acid (100 µM, 3MPA, Toronto Research Chemicals, n = 8), or vehicle (MilliQ water, n = 8). We use a glass micropipette adapted to a Hamilton syringe, this procedure is minimally invasive and there was no mortality associated. Tissues for immunohistochemistry analysis were obtained at P3, by killing the animals by anesthetic overdose followed by a transcardial perfusion with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer, pH 7.3. The mouse brains were post-fixed for 8–12 h, cryoprotected, and kept frozen. Coronal sections of 40 μm thickness were collected in a cryoprotective solution and stored at −30°C until further use.
Thymidine analog 5-bromo-2-deoxyuridine (BrdU; Sigma-Aldrich) was injected intraperitoneally (5 mg/10 g body weight) 3 h after lactate or vehicle injection to P0 animals. At P3, animals were perfused and processed as described above. BrdU incorporation was analyzed by immunohistochemistry (rat-anti-BrdU, 1:1000; Abcam, UK). The cell types that incorporated BrdU and their progeny were identified by double immunohistochemistry with antibodies against BrdU and Ki67, Sox2, Tbr2, DCX, or BLbP.
Neurons were obtained from embryonic brains as described elsewhere (Ortega and Alcantara 2010). Briefly, a time-pregnant mouse was killed by cervical dislocation and the embryos extracted at embryonic day 16 (E16).
Cerebral cortices were dissected free of the meninges in a solution of PBS with 0.6% glucose (Sigma-Aldrich) and 0.3% bovine serum albumin (Sigma-Aldrich) and then digested with trypsin (Biological Industries, Israel) and DNAse I (Sigma-Aldrich) for 10 min at 37°C. The tissue was mechanically dissociated, centrifuged, and resuspended in CO2-equilibrated Dulbecco's modified Eagle's medium supplemented with 10% normal horse serum (NHS), 1% pen-strep, 0.5 mM l-glutamine, and 5.8 µL NaHCO3 (Sigma-Aldrich)/mL. The cell suspension was preplated at 37°C. After 30 min, the supernatant was collected, centrifuged (1000 g for 5 min), resuspended in NB neuronal culture medium (1% NHS, 1% pen-strep, 0.5 mM l-glutamine, 22 μM glutamic acid (Sigma-Aldrich), 2% B27 (Gibco, USA), and 5.8 µL NaHCO3/mL), and plated at a density of 2.5 × 105 cells/cm2 directly on tissue culture plates coated with poly-d-lysine (Sigma-Aldrich). After 24 h, the medium was replaced with serum-free neuronal culture medium (1% pen-strep, 0.5 mM l-glutamine, 2% B27, 5.8 µL NaHCO3/mL) during 1 days in vitro (div), 3 div, and 5 div. Under these conditions, we obtained a neuron-enriched culture in which about 10% of the cells were glial cells.
The samples were fixed in 4% PFA for 15 min at room temperature (RT) for immunofluorescence, protein extraction and western blot analysis, and ATP measurements. Media were collected to measure lactate and glucose concentrations.
For western blot analysis, protein extracts were obtained from primary cultures. Total cell proteins were separated by SDS-PAGE and electrotransferred from the gel to a nitrocellulose membrane (Bio-Rad). The membranes were blocked and then incubated first with primary antibodies overnight at 4°C, and then with their corresponding secondary HRP-conjugated antibodies (1:3000; Santa Cruz Biotechnology, USA). Protein signals were detected using the ECL chemiluminescent system (Amersham, GE Healthcare, PA, USA). Densitometry analysis, standardized to F-actin as a control for protein loading, was performed using ImageJ software. For quantification, triplicate samples were analyzed and at least 2 different experiments were conducted.
For immunofluorescence, fixed samples (4% PFA for 15 min at RT) were incubated with primary antibodies and appropriate Alexa 488 or Alexa 555 secondary antibodies (1:500, Molecular Probes, USA). The following primary antibodies were used: goat anti-actin (cytoskeleton marker, 1:2000, Santa Cruz Biotechnology), mouse anti-Tuj-1 (neuronal marker 1:10000, Covance), rabbit anti-Tbr2 (neuronal restricted progenitor cells marker, 1:500, Abcam), rabbit anti-Ki67 (proliferation marker, 1:500, Abcam), rabbit anti-Sox2 (NSC/nonrestricted progenitor marker, 1:1000, Abcam) goat anti-MCT2 (neuronal monocarboxylate transporter marker 1:500, Santa Cruz Biotechnology), rabbit anti-PEPCK-M (mitochondrial phosphoenolpyruvate carboxykinase, 1:1000, Abcam), goat anti-GPR81 (G-protein-coupled receptor 1:500, Santa Cruz Biotechnology), rabbit anti-phospho-AMPKα (adenosine monophosphate-activated protein kinase marker 1:1000, Cell Signaling), rabbit anti-Sirt1 (NAD-dependent deacetylase sirtuin-1 marker 1:1000, Upstate), rabbit anti-phospho-PDH-E1α (phosphorylated pyruvate dehydrogenase E1-alpha, 1:5000, Millipore), rat-anti-BrdU (cycling cells marker, 1:1000, Abcam), rabbit anti-DCX (doublecortin, immature neuronal marker, 1:6000, Abcam), rabbit anti-BLbP (glial marker, 1:3000, Chemicon) rabbit anti-AKT (serine/threonine-specific protein kinase, 1:1000, Millipore), rabbit anti-laminin-EHS (1:500, Sigma-Aldrich, S9393), rabbit anti-caspase-3 (apoptosis marker 1:100, Santa Cruz Biotechnology), MitoTracker (250 nM, mitochondria marker, Invitrogen), and To-Pro-3 iodide (Topro) (nuclear stain, 1:500, Molecular Probes). Finally, the preparations were coverslipped with Mowiol (Calbiochem) for imaging.
E16 primary neuronal cultures treated with l- or d-lactic acid (Sigma-Aldrich) were cultured in Neurobasal medium supplemented with 25 mM glucose (glucose medium) or with 25 mM glucose and 4 mM lactic acid ((glucose + lactate medium), or with 4 mM lactic acid only (glucose-free lactate medium). Neurons were cultured for 24 h in the above media supplemented with 1% NHS, 1% pen-strep, 0.5 mM l-glutamine, 22 μM glutamic acid, 1× B27, and 5.8 μL NaHCO3/mL. The medium was then replaced with the same medium but serum-free and the cultures incubated for 4 more days.
In some experiments, lactate or glucose medium was replaced after 5 div with glucose or lactate medium for 1 h (short incubation) or 2 days (long incubation). All experiments were carried out at least 3 times and triplicate samples from each were analyzed by western blot, Inmunofluorescence (as a measure of cell death), or spectrophotometrically (to determine the lactate content of the medium).
ATP in cell extracts was measured using a luciferin/luciferase assay (BioVision). Cell extracts were prepared in a single-step boiling water procedure (Yang et al. 2002). In brief, the cells were washed twice with PBS and then suspended in 200 µL of boiling water. The cell suspension was centrifuged (12000g, 5 min, 4°C) and the resulting supernatant was assayed. Ten µL of the cell suspension was mixed with 10 µL of luciferin/luciferase reagent and 80 µL of deionized water. Light emission was measured on a TD 20/20 luminometer (Turner Designs, Houston, USA). A standard ATP curve was obtained by serial dilutions of 10 mM ATP (Sigma-Aldrich). The amount of ATP in the cell extracts was normalized to the cellular protein concentration.
Oxidative stress was measured using the cell-permeable fluorogenic probe CellROX Green (Invitrogen) as recommended by the manufacturer. In its reduced state CellROX Green is weakly fluorescent whereas upon oxidation by reactive oxygen species it binds to DNA and exhibits bright green fluorescence (excitation/emission: 485/520 nm). Cells assayed by this method were cultured in lactate medium as described above; after 5 div they were shifted to glucose or lactate medium for either 1 h (short incubation) or 2 days (long incubation). The CellROX Green reagent was added at final concentration of 5 µM and the cultures were incubated at 37°C for 30 min. The cells were fixed with 4% PFA, mounted with Mowiol, and their fluorescence was then analyzed using ImageJ software (National Institutes of Health, USA).
Experiments testing the effects of MCT2 inhibition or the lactate receptor agonist GPR81 were carried out as follows: E16 primary neuronal cultures were treated with l- or d-lactic acid (Sigma-Aldrich) and cultured in 3 different media: NB medium (with 25 mM glucose, glucose medium), NB medium supplemented with 4 mM lactic acid (glucose + lactate medium) or glucose-free NeurobasalA medium (NBA) supplemented with 4 mM lactic acid (lactate medium). Neurons were cultured for 24 h in these media, also containing 1% NHS, 1% pen-strep, 0.5 mM l-glutamine, 22 μM glutamic acid, 1× B27, and 5.8 μL NaHCO3/mL. The medium was then changed to serum-free formulations containing 1% pen-strep, 0.5 mM l-glutamine, 1× B27, 5.8 μL NaHCO3/mL and the presence or absence of the MCT2 inhibitor AR-C155858 (100 nM, Adooq) or the GPR81 agonist, 3,5-dihydroxybenzoic acid (1 mM, 3,5-DHBA, Santa Cruz) for 4 more days.
In experiments testing the effects of PEPCK inhibition, the cells were cultured as described above, except that the serum-free medium contained either DMSO (1:2000, Sigma-Aldrich) for control conditions or the PEPCK inhibitor 3-mercaptopicolinic acid (100 µM, 3MPA, Toronto Research Chemicals). The cells were then cultured for 4 more days.
The concentration of l-lactate was determined using an enzymatic reaction based on the oxidation of l-lactate to pyruvate by lactate dehydrogenase (5 mg of the enzyme (Roche)/mL, 550 U/mg) in the presence of NAD (Sigma-Aldrich). In this assay, the amount of NADH produced in the reaction is proportional to the amount of l-lactate in the samples. With this enzymatic system, d-lactate is not detected.
All experiments were carried out at least 3 times and in triplicate samples. The latter were diluted 1:20 with reaction mix [0.3 M hydrazine sulfate (Merck) and 0.87 M glycine (AppliChem), pH 9.5; 2.5 M NAD+ (Sigma-Aldrich), 0.19 M EDTA (Merck)]. Lactate dehydrogenase was added at a final concentration of 6.9 U/mL. The NADH concentration was determined by using the Fluostar Optima BMG Labtech system to measure absorbance (340 nm) and fluorescence (excitation 340 nm/emission 460 nm) 0 and 20 min after the start of the reaction. The endpoint of the reaction was set at 20 min and the corresponding values were used in the calculations. Different concentrations of sodium l-lactate (Sigma-Aldrich) served as the standard.
Imaging and Cell Analysis
Digital images were taken throughout the study using a software-controlled digital camera. Fluorescent preparations were visualized and micrographs were captured with either a Leica TCS-SL Spectral confocal microscope (Leica Microsystems, Mannheim, Germany) or a Nikon Eclipse 800 light microscope (Nikon, Tokyo, Japan). The images were assembled in Adobe Photoshop (v. 7.0), with identical adjustments for contrast, brightness, and color balance to obtain optimum visual reproductions of the data. Morphometric and quantitative analyses were performed using ImageJ software (National Institutes of Health, USA).
Cell counts were expressed as mean cells/mm2 ± standard deviation. The values are the average of 3 replicates of at least 2 different experiments. Statistical analysis was performed using the Statgraphic-plus software and GraphPad Prism. One-way ANOVA and Fisher's least significant difference (LSD) procedure were used to discriminate between the means.
Effect of Lactate on Neuronal Progenitor Survival and Differentiation
To explore the influence of l-lactate on the self-renewal and differentiation of neuronal progenitors in vitro, primary embryonic neuronal cultures were grown in serum-free culture medium with glucose (25 mM glucose, glucose medium), glucose and l-lactate (25 mM glucose + 4 mM l-lactic acid, glucose + lactate medium), or l-lactate only (4 mM l-lactic acid, lactate medium) (see Material and Methods). Under these 3 conditions, initial cell adherence was the same while in the presence of lactate, glial content was residual or mostly absent (Alvarez et al. 2013). After 1 day in vitro (1 div) (Fig. 1A) similar counts were obtained for both the total number of cells (glucose: 274 ± 43, glucose + lactate: 243 ± 46, and lactate: 246 ± 36) and the number of Ki67+ cycling progenitors (glucose: 38 ± 9, glucose + lactate: 44 ± 14, and lactate: 55 ± 14) (Fig. 1C). After 5 div, the neuronal cultures showed signs of differentiation, and neurons identified by staining with the neuronal marker Tuj-1 exhibited well-developed neurites, particularly in glucose medium (Fig. 1B). At this time point, the total cell numbers in the glucose (375 ± 47, P < 0.01) and glucose + lactate (297 ± 40, P < 0.05) cultures were significantly higher than at 1 div while in lactate medium there was no change (247 ± 27). A similar comparison of Ki67+ progenitors (Fig. 1D) showed fewer numbers of these cells in glucose medium (4 ± 3, P < 0.01), while the number remained constant in glucose + lactate medium (57 ± 19) and had significantly increased in lactate medium (79 ± 18, P < 0.01). As post-mitotic neurons cannot reenter the cell cycle, the increase in cell number in 5 div embryonic neuronal cultures can be explained by the division of preexisting Ki67+ progenitors at 1 div. Therefore, the increase in cell number and the depletion of progenitors after 5 div in glucose medium is compatible with 2 rounds of progenitor division; first an asymmetric division giving rise to a progenitor and a differentiated cell, and then a symmetric division of the progenitor giving 2 differentiated cells. The observation in glucose + lactate medium of an increased number of cells together with progenitor maintenance is compatible with one round of asymmetric progenitor division giving rise to a self-renewing progenitor and a differentiated cell; and finally, the increase in progenitors but not in the total cell number in lactate medium suggests self-renewing symmetric divisions and an enrichment in progenitor cells compared with the initial pool of terminally differentiated cells.
Based on these observations, we speculated that glucose promotes the differentiation of neuronal progenitors while lactate is required for their self-renewal. To test this hypothesis, neuronal cultures were grown in lactate medium for 5 div and then for 2 more days either in the same medium or in glucose medium (Supplementary Fig. 1A). After the 7 days, there were significantly fewer Ki67+ progenitors in glucose than in lactate medium (10 ± 1 and 63 ± 5 respectively, P < 0.001, Supplementary Fig. 1B), while the total number of cells was significantly higher in glucose than in lactate cultures (220 ± 41 and 170 ± 23 respectively, P < 0.001, Supplementary Fig. 1C). These results suggested that glucose medium induced one terminal division of Ki67 progenitors whereas in lactate medium these cells remained mostly quiescent. When the opposite experiment was done, i.e., neuronal cultures were grown in glucose medium for 5 div and then for 2 more days either in the same medium or in lactate medium (Supplementary Fig. 1D), there were no changes either in the total number of cells (260 ± 34 glucose; 228 ± 44 lactate) or in the number of Ki67+ progenitors (2 ± 2 glucose; 1 ± 1 lactate) (Supplementary Fig. 1E,F), indicating that lactate cannot reprogram the terminal progenitor differentiation induced by glucose.
We then examined whether lactate induced the formation of reactive oxygen species, leading to oxidative stress and cell death and therefore the observed differences in cell number. Neuronal cultures were grown in lactate medium for 5 div and then in the same medium or in glucose medium. After either 1 h or 2 days, the cells were treated with CellROX, an indicator of oxidative stress, for 30 min before fixation. After 1 h, CellROX fluorescence was higher in neuronal cultures grown in glucose than in lactate medium (21 ± 3 and 16 ± 2 cells respectively, P < 0.001). After 2 days, however, CellROX intensity decreased and the differences disappeared (glucose: 10 ± 1 cells, lactate: 10 ± 5 cells) (Supplementary Fig. 1G,H). As an indicator of cell death, the number of caspase-3+ cells was analyzed under the same experimental conditions. The percentage of dead cells was very low in either medium although at 1 h it was slightly higher in the lactate than in the glucose-containing cultures (glucose: 2 ± 2, lactate: 5 ± 2, P < 0.05). These differences were no longer observed after 2 div (glucose: 3 ± 3, lactate: 2 ± 3) (Supplementary Fig. 1I,J).
Effect of Lactate on Lactate-signaling and Metabolism
Next, we analyzed whether the presence of lactate in the medium affected the expression of lactate-related machinery, i.e., the high-affinity proton-linked monocarboxylate transporter MCT2 (Halestrap and Wilson 2012), the G-protein-coupled lactate receptor, GPR81 (Lauritzen et al. 2013), and the mitochondrial enzyme PEPCK-M, required for the anabolic use of lactate in the liver (Mendez-Lucas et al. 2013). After 5 div in glucose medium, MCT2 was detected at low levels in neuronal membranes. In lactate media MCT2 was abundantly expressed in membranes and cytosol of neurons, Ki67+ and Tbr2+ progenitors (Fig. 2A,B). Tbr2 transcription factor is required for the conversion of radial glia into neuronal restricted progenitors (Sessa et al. 2008), that is the principal type of progenitor induced by lactate treatments (Alvarez et al. 2013). In contrast, GPR81 (Fig. 2C) was expressed in neuronal membranes in glucose medium and by some but not all Ki67+ progenitors (not shown) and Tbr2+ neuronal progenitors in lactate conditions. Similarly, in glucose medium, PEPCK-M was detected at low levels, with a punctate distribution in the neuronal soma, whereas in the presence of lactate both neurons and progenitors expressed very high levels of the enzyme (Fig. 2D). Quantitative analysis by western blot and densitometry (Fig. 2E,F) corroborated the significant lactate-related increases of these 4 proteins, MCT2, GPR81 and PEPCK-M, and Tbr2. Moreover, PEPCK-M and Tbr2 expression was significantly higher in lactate than in glucose + lactate medium. The absence of Sox2+ progenitors (Fig. 2E,F) and astrocytes (Alvarez et al. 2013) in lactate cultures suggested that the Ki67+ progenitors were mostly Tbr2+ neuronal progenitors and that lactate and PEPCK-M activity allowed the avoidance of differentiation, perhaps by promoting the maintenance of self-renewal.
In a pharmacological approach we sought to answer the question whether neuronal survival and progenitor maintenance required lactate intake, either through monocarboxylate transporters, activation of the GPR81 lactate receptor, PEPCK-M catalytic activity, or a combination of all 3. Thus, neuronal cultures were grown for 5 div in the presence or absence of 100 nM of the MCT1/2 inhibitor AR-C155858 beginning at 1 div. When added to glucose + lactate medium, the inhibitor induced Ki67+ progenitor depletion (22 ± 5 and 0 ± 1 respectively, P < 0,001), without affecting neuronal survival (211 ± 51 and 240 ± 24, respectively), whereas when added to lactate medium it induced massive cell death (159 ± 35 and 0, respectively) (Fig. 3A). In a second set of experiments, 3,5-DHBA, a selective agonist of GPR81, was added to glucose medium or to glucose/lactate-free medium beginning at 1 div. After 5 div in glucose medium, the agonist did not modify either the total number of cells (glucose: 255 ± 46, glucose + agonist: 216 ± 79) or the number of Ki67+ progenitors (glucose: 0 ± 0, glucose + agonist: 0 ± 0), nor did it promote cell survival in glucose/lactate-free medium (Fig. 3B). Moreover, the effects of l-lactate on neuronal cultures were not mimicked by d-lactate (Supplementary Fig. 2).
To specifically examine the potential role of PEPCK in shuttling carbons from lactate through cataplerosis into various metabolic pools, we treated cultures during 5 div with 3-mercaptopicolinic acid (3MPA), a well-known inhibitor of PEPCK activity, or with DMSO as the control (Fig. 3C,D). In glucose medium, PEPCK inhibition did not affect the total number of cells (256 ± 50 DMSO; 252 ± 54 3MPA) or the number of Ki67+ progenitors (1 ± 2 DMSO; 1 ± 2 3MPA). However, in glucose + lactate or lactate medium, the lack of PEPCK activity corresponded with a significant reduction in the total number of cells (from 204 ± 47 to 150 ± 46 in glucose + lactate and from 121 ± 50 to 79 ± 34 in lactate, P < 0.01) as well as the complete depletion of Ki67+ progenitors (from 12 ± 5 glucose + lactate, 15 ± 6 lactate to 0, P < 0.01), indicating that in the presence of lactate, PEPCK-M inhibition leads to progenitor cell death. Consistent with this finding, lactate consumption decreased in the presence of 3MPA (from 0.7 ± 0.1 to 0.2 mM in glucose + lactate, and from1.7 ± 0.1 to 1.5 ± 0.2 mM in lactate, P < 0.05) (Fig. 3E).
Taken together these data suggest stereospecific effects of l-lactate on neuronal survival and progenitor self-renewal. These effects are apparently mediated by the cellular entry of l-lactate through MCT1/2 receptors and its metabolism at the TCA junction, represented by PEPCK-M catalytic activity.
Lactate-associated Changes in the Metabolic Profile of Neuroprogenitor Cells
To investigate changes in the metabolic profile of self-renewing progenitor cells in the presence of lactate several key elements of metabolic regulation were analyzed. Differences in mitochondrial distribution were visualized by epifluorescence using MitoTracker, a cell-permeable mitochondrial marker. In neuronal cultures grown in glucose, glucose + lactate, or lactate medium for 5 div well-developed neurons stained with the neuronal marker Tuj-1 contained fewer mitochondria than young neurons and progenitors (Tuj-1 negative), which showed intense mitochondrial labeling (Fig. 4A). Consistent with progenitor enrichment, mitochondrial staining was highest in lactate medium, while there was no medium-related difference in ATP levels (Fig. 4B). Western blot and densitometry analysis showed slightly higher levels of phosphorylated AMPK (AMPK-P) in glucose + lactate medium, without concomitant changes in NAD-dependent deacetylase sirtuin-1 (Sirt1) or phosphopyruvate dehydrogenase (PDH)-E1−α [PDH E1α-P(Ser293)], whereas AMPK-P, Sirt1, and phospho-PDH-E1−α levels were much lower in lactate condition (Fig. 4C,D) than in either of the glucose-containing media, indicating striking increased in the energy charge, redox potential, and mitochondrial metabolism of lactate through PDH. These data are consistent with a dramatic shift to oxidative metabolism as lactate becomes the main energy and carbon source in the cell. The metabolic profile described is a newly identified hallmark of neuron restricted progenitor cells and contrasts with the glycolytic profile of glucose restricted cells (mainly the glial component).
Lactate restriction of neuronal progenitors implies that lactate metabolism also plays a crucial role in sustaining the increase in biomass that occurs during the cell cycle and in the ECM production characteristic of these cells. In this context, PEPCK-M is the only pathway that can export carbons from lactate and other TCA cycle intermediates (i.e., glutamine) into the triose and hexose pools, through PEP synthesis (Mendez-Lucas et al. 2013). Therefore, we evaluated whether the presence of lactate and the inhibition of PEPCK-M activity using 3MPA altered the synthesis of laminin, a highly glycosylated ECM protein that serves as a marker of synthetic activity in neuronal progenitors (Kazanis et al. 2010). Indeed, lactate significantly increased the laminin content of neuronal cultures (Fig. 4E,F), mainly associated with progenitor cells (Supplementary Fig. 3). In the presence of lactate and 3MPA there was a substantial decrease in the laminin content as well as in Tbr2+ neuronal progenitors as shown by western blot (Fig. 4G,H). This result suggested that the PEPCK-M pathway is relevant to link the mitochondrial and glycolytic pools of building blocks necessary to sustain biosynthetic processes.
Requirement of Lactate for Neural Progenitors in vivo
We tested whether progenitor maintenance in the cerebral cortex was effectively mediated by l-lactate and PEPCK-M catalytic activity. In this in vivo experiment, 2 µL of either l-lactate (5 mM, lactate), the PEPCK-M inhibitor 3MPA (100 µM, 3MPA), MCT1/2 inhibitor AR-C155858 (100 nM, MCT inhib), or vehicle (control) was injected into the lateral ventricle of newborn (postnatal day 0, P0) mice (Fig. 5A). The effects on PEPCK-M expression and in the number of Ki67+ cycling cells, Sox2+ NSC/nonrestricted progenitors, and Tbr2+ neuronal progenitors were analyzed in the ventricular/subventricular and intermediate zones (VZ/SVZ, IZ) at P3 (Fig. 5). The results showed the PEPCK-M expression in the germinal VZ/SVZ of lactate-injected brains (Fig. 5B) together with significant increases in the number of Ki67+ progenitors (vehicle: 83 ± 14, lactate: 129 ± 37, MCT inhib: 73 ± 11, 3MPA: 45 ± 13; Fig. 5C,D), Sox2+ NSC/progenitors (vehicle: 94 ± 13, lactate: 125 ± 33, MCT inhib: 80 ± 22, 3MPA: 57 ± 9; Fig. 5E,F), and Tbr2+ neuronal progenitors (vehicle: 57 ± 10, lactate: 72 ± 12, MCT inhib: 34 ± 12, 3MPA: 23 ± 8; Fig. 5G,H). Conversely, the numbers of all these cells decreased significantly in 3MPA while only the number of Tbr2+ neuronal progenitors was significantly reduced by the MCT inhibitor. Moreover, compared with the vehicle control, laminin expression in the germinal VZ/SVZ dramatically increased in lactate-injected animals but decreased in 3MPA and MCT inhibitor-injected animals (Fig. 5I). Taken together these data provide in vivo corroboration of the lactate dependence of neural progenitors.
Finally, we examined lactate-mediated changes in progenitor fate using BrdU, a marker for DNA replication. BrdU was injected 3 h after lactate or vehicle (P0) administration and incorporation into different cell types was analyzed at P3 (Fig. 6A). The total number of BrdU labeled cells was significantly higher in lactate injected animals than in controls (control: 98 ± 31, lactate: 122 ± 38, P < 0.05). In lactate injected animals there was a significant increase in the number of BrdU+ progenitors expressing Ki67 (vehicle: 39 ± 11, lactate: 54 ± 12; Fig. 6B,C), Sox2 (vehicle: 27 ± 11, lactate: 38 ± 7; Fig. 6D,E), and Tbr2 (vehicle: 4 ± 1, lactate: 7 ± 1,; Fig. 6F,G). However, the number of double labeled BrdU and DCX+ immature neurons (vehicle: 18 ± 7, lactate: 15 ± 3; Fig. 6H,I) or BLbP+ glial cells (vehicle: 11 ± 3, lactate: 12 ± 4,; Fig. 6J,K) were similar in both treatments. The consequence is a net increase in the total number of progenitors 3 days after lactate injection without significant changes in the number of differentiated neurons and glial cells generated in this period. These data might indicate that neonatal lactate maintains progenitor self-renewal in vivo, corroborating in vitro results.
Metabolism is an important indicator of cell function, since it shifts together with differentiation, growth, or anabolic capacities. For example, undifferentiated stem cells preferentially rely on glycolysis whereas differentiated cells up-regulate oxidative metabolism to support their anabolic or biological potential (Facucho-Oliveira and St John 2009; Ivanovic 2009; Simsek et al. 2010). In the developing brain, NSC/radial glia progenitors generate the various differentiated cell types (Rowitch and Kriegstein 2010). Several studies have shown that the highly hypoxic environment characteristic of the stem cell and NSC niche favors glycolysis and lactate production at sites of unrestricted progenitor cell proliferation (Mohyeldin et al. 2010). As NSCs mature, their specific lineage restriction is likely to be accompanied by a shift in their metabolic needs (Stubbs et al. 2009; De Filippis and Delia 2011; Goldman and Chen 2011). This shift might explain the role of l-lactate, previously shown to be a critical substrate of the developing brain (Medina and Tabernero 2005; Rinholm et al. 2011). Indeed, our results provide evidence of a lactate-restricted metabolic requirement to fulfill the biosynthetic needs of neuronal restricted progenitors (summarized in Fig. 7).
Aerobic glycolysis is a functional marker of the specialized phenotype of dividing cells, whether they arise from physiologic tissue components (i.e., stem cells or endothelial cells) or after pathologic dedifferentiation (tumor cells) (De Bock et al. 2013; Suda et al. 2011; Takubo et al. 2013). This seemingly inefficient, partial oxidation of glucose nonetheless provides both a rapid source of energy and the anabolic building blocks to support cell division, but it reduces the cellular capacity to integrate into a functional, complex tissue. The same can be said for NSC metabolism (Yamasaki et al. 2001; Rafalski and Brunet 2011) and of radial glia progenitors and astrocytes, which are also essentially glycolytic (Tsacopoulos and Magistretti 1996; Yamasaki et al. 2001). In contrast, in this work we showed that neuronal restricted progenitors have an oxidative metabolic profile that is entirely dependent on lactate as a carbon source. In the presence of glucose, lactate withdrawal induces progenitor differentiation to a neuronal fate.
Consistently, key metabolic markers of oxidative metabolism, including increased mitochondrial content and high PDH activity (reduced phosphorylation), are classically found in differentiated cells with a high anabolic capacity, such as liver, heart and slow-twitch muscle, and pancreatic β-cells. Terminal oxidative phosphorylation of pyruvate, whether from glucose or lactate sources, ensures the appropriate redox potential and energy charge needed to sustain biosynthesis at a high level of efficiency; while ensuring minimal rates of cell division. The progenitors examined in this study underwent one or 2 rounds of cell division during their 5 days in culture, even though in all cases their nuclei were Ki67+. The cellular redox state contributes to the determination of cell fate. Thus, for example, increases in Sirt1 activity under oxidizing conditions induce neural progenitors to adopt a glial fate (Prozorovski et al. 2008), while under basal nonoxidizing conditions these progenitors differentiate along a neuronal line (Hisahara et al. 2008). Our data showed low-level Sirt1 expression, consistent with an increased redox potential, in parallel with a restriction to a neuronal progenitor fate. Also, diminished AMP-K phosphorylation suggested that when lactate is the main carbon source both redox potential and energy charges are increased, in line with the requirements of a mainly anabolic metabolism. Hence, we propose that Tbr2+ neuronal restricted progenitors have a role in maintaining the ECM of the neurogenic niche and in the secretion of factors necessary for the neurogenic environment. Laminin is an integral component of the ECM of neurogenic niches (Kazanis et al. 2010) and, as shown in the present work and in a previous study (Alvarez et al. 2014), it is secreted by neural cells and progenitors exposed to lactate. Indeed, we found that laminin production was sensitive to the PEPCK inhibitor 3MPA, suggesting that both PEP and TCA flux are necessary to support biosynthetic processes, such as laminin production and glycosylation. PEPCK activity is present in the cytosol (PEPCK-C) and in mitochondria (PEPCK-M), encoded by 2 different nuclear genes, PCK1 and PCK2, respectively. In neural progenitors, only the mitochondrial isoform, PEPCK-M, was detected, with no measurable cytosolic protein found (data not shown). PEPCK-M catalyzes the GTP-dependent decarboxylation of mitochondrial oxaloacetate to produce PEP, which is then exported into the cytosol where it feeds the reverse glycolytic pathway (Stark et al. 2009; Mendez-Lucas et al. 2013). PEPCK-M has important advantages over PEPCK-C, since the shuttling of malate is not required and enzyme activity is coupled to TCA cycle flux through the recycling of GTP produced in the succinyl-CoA synthase reaction (Stark et al. 2009). This pathway also shuttles carbons from lactate into the triose-phosphate intermediate pool for the synthesis of serine/glycine or glycerol-3P, or into hexose-phosphate intermediates for protein glycosylation, in the absence of oxidative phosphorylation (Nye et al. 2008; Yang et al. 2009; Kalhan and Hanson 2012). It is noteworthy that PEPCK-M is the only known pathway that can communicate mitochondrial carbon intermediates directly to the glycolytic pool, because of the irreversible nature of the pyruvate kinase step.
Lactate injection in vivo produced similar results, and confirmed that lactate increased the progenitor BrdU-traced pool. As described in vitro, lactate induced PEPCK-M expression, laminin synthesis and Tbr2+ neuronal restricted progenitors in vivo, and the inhibition of lactate machinery (MCT1/2 lactate transporter or PEPCK-M activity) significantly reduced the number of neuronal restricted progenitors. In vitro, Sox2+ cells with NSC characteristics were rarely seen in our neuronal cultures, however in vivo lactate injection induced an increase in Sox2+ progenitors and the inhibition of PEPCK-M activity significantly reduced their number. These data are in accordance with previous results demonstrating the requirement of the NSC niche for specific topographical cues. Thus, while Tbr2+ progenitors are induced by a flat substrate of lactate-releasing polylactic acid (PLA), Sox2+ progenitors/NSC are only induced when the same PLA substrate is molded as aligned nanofibers, a 3D organization that reproduces the aligned palisade of embryonic radial glia (Alvarez et al. 2013; 2014).
On the other hand, MCT1/2 inhibitor in vivo only affected the number of neuronal progenitors, as compared with the broad progenitor reduction observed after PEPCK-M activity blockage. The complexity of lactate transport across membranes in vivo might explain these data as it is dependent on MCT1-4 transporters expression and modulated by intra and extracellular concentration, pH and proton gradient, and in addition lactate can enter the cell trough other ways including gap junctions (Hertz et al. 2014). MCT2 is expressed in the neuronal lineage while MCT1 and 4 are found in glial cells (Debernardi et al. 2003).
Taken together, our results support the strict dependence of neuronal restricted progenitors on lactate metabolism and on the provision of ECM components of the neurogenic niche by the anabolic activity of PEPCK-M. In the in vivo niche, other physical and/or biochemical cues might cooperate with lactate to maintain different progenitor populations. Further experiments will be required to determine the impact of lactate metabolism modulation on the neuronal composition and functional organization of the cerebral cortex.
This study was supported in part by grants from Spain's Ministerio de Economía y Competitividad (MAT2011-29778-C02-02) and (BFU2012-37177), cofinanced by the European Regional Development Fund to S.A and J.C.P respectively; and from Generalitat de Catalunya 2009 SGR719 to S.A.
Z.A acknowledges the fellowship from IBEC 10-2009-01 and P.H was supported by a fellowship from the Ministerio de Educación y Ciencia (FPU). We are grateful to Wendy Ran for editorial assistance, to J.A. Ortega for the drawing in Figure 5A and B. Torrejón from the Scientific-Technical Services UB (Campus Bellvitge) for technical support in confocal microscopy. Conflict of Interest: None declared.