Abstract
Cellular metabolism is important for adult neural stem/progenitor cell
(NSPC) behavior. However, its role in the transition from quiescence to
proliferation is not fully understood. We here show that the
mitochondrial pyruvate carrier (MPC) plays a crucial and unexpected
part in this process. MPC transports pyruvate into mitochondria,
linking cytosolic glycolysis to mitochondrial tricarboxylic acid cycle
and oxidative phosphorylation. Despite its metabolic key function, the
role of MPC in NSPCs has not been addressed. We show that quiescent
NSPCs have an active mitochondrial metabolism and express high levels
of MPC. Pharmacological MPC inhibition increases aspartate and triggers
NSPC activation. Furthermore, genetic Mpc1 ablation in vitro and in
vivo also activates NSPCs, which differentiate into mature neurons,
leading to overall increased hippocampal neurogenesis in adult and aged
mice. These findings highlight the importance of metabolism for NSPC
regulation and identify an important pathway through which
mitochondrial pyruvate import controls NSPC quiescence and activation.
__________________________________________________________________
Inhibition of pyruvate entry into mitochondria promotes the activation
of adult neural stem cells and increases neurogenesis.
INTRODUCTION
Stem cells must maintain a tight balance between quiescence,
proliferation, and differentiation to sustain lifelong tissue
regeneration and maintenance. This is also the case for adult neural
stem/progenitor cells (NSPCs), which form newborn neurons throughout
life ([34]1). NSPCs are primarily quiescent in adulthood but can
proliferate upon intrinsic and extrinsic stimuli ([35]2). Although NSPC
activation is critical for proper neurogenesis, the underlying
mechanisms are still not fully understood.
Cellular metabolism has been shown to determine the activity state of
stem cells ([36]3, [37]4), and metabolic features appear similar among
different tissue-specific adult stem cells. In general, stem cells are
rather glycolytic to support synthesis of cellular building blocks to
sustain cell growth, while during differentiation, their metabolic
profile shifts toward oxidative metabolism to generate adenosine
triphosphate ([38]5–[39]13). Such a metabolic shift seems also
important for NSPCs. Single-cell RNA sequencing (scRNA-seq) studies
found decreased expression of glycolytic genes and an up-regulated
expression of genes involved in oxidative phosphorylation (OXPHOS) at
early stages of fate transition in adult NSPCs ([40]14, [41]15). These
findings are supported by metabolic analyses of NSPCs under
differentiation in vitro ([42]16, [43]17). However, recent findings
suggest that when NSPCs are in a quiescent state, their metabolism
might be substantially different from glycolytic, proliferating NSPCs:
Quiescent NSPCs have high levels of mitochondrial fatty acid
β-oxidation (FAO) and express many proteins involved in diverse aspects
of mitochondrial metabolism ([44]18, [45]19). Furthermore, mitochondria
are abundant in NSPCs, and their dynamics affects self-renewal and fate
choice ([46]20, [47]21). Thus, mitochondrial metabolism might play a
more important role for NSPC quiescence than previously anticipated.
A shift from a glycolytic to more oxidative metabolism very often
requires a redirection of pyruvate, the end product of glycolysis, from
lactate production toward mitochondrial oxidation in the tricarboxylic
acid (TCA) cycle. The mitochondrial pyruvate carrier (MPC), a
heterodimer of MPC1 and MPC2, is required for this transport
([48]22–[49]24). Absence of one of the two proteins leads to a loss of
pyruvate transport and has a profound impact on the metabolic state of
the cells ([50]25, [51]26). Despite its key role in linking glycolysis
and mitochondrial metabolism, it remains unknown whether MPC plays a
regulatory role in NSPC behavior.
We here used pharmacological MPC inhibition and genetic deletion of
Mpc1 to assess whether a disruption of pyruvate import into
mitochondria would affect NSPC maintenance, activation, and
differentiation. Unexpectedly, we found that quiescent NSPCs express
high levels of MPC and require pyruvate import into mitochondria for
the maintenance of quiescence. Inhibition of MPC triggers their
activation by increasing the intracellular pool of aspartate despite a
substantial decrease of TCA cycle intermediates. Furthermore,
conditional MPC1-knockout (cKO)NSPCs are able to differentiate into
mature neurons, indicating a high metabolic flexibility, which allows
these cells to adapt their metabolism according to substrate
availability. We further show that this increased activation and
undisturbed differentiation of MPC1-cKO NSPCs leads to an overall
increase in neurogenesis in adult and middle-aged mice.
RESULTS
MPC is dynamically regulated with activity state, and its transport function
is required for NSPC quiescence
To determine the role of MPC in NSPCs, we first analyzed the expression
of Mpc1 in existing RNA-seq databases. We found that Mpc1 is expressed
in NSPCs and in other cell types in the dentate gyrus (DG) of adult
mice ([52]Fig. 1A) ([53]27). Available images from DG sections of
MPC1–green fluorescent protein (GFP) reporter mice ([54]www.gensat.org)
further showed expression of GFP in the subgranular zone of the DG
(fig. S1A).
Fig. 1. MPC is dynamically regulated with activity state, and its transport
function is required for NSPC quiescence.
[55]Fig. 1.
[56]Open in a new tab
(A) t-distributed stochastic neighbor embedding visualization of Mpc1
expression in the DG of adult mice ([57]linnarsonnlab.org/dentate).
NSPCs are shown as radial glia (RG) and radial glia–like cells (RGL).
(B) Violin plots of Mpc1 expression [transcripts per million (TPM)] in
quiescent NSPCs, activated NSPCs and intermediate progenitor cells,
data queried from ([58]27). Red line represents mean. *P < 0.05
[one-way analysis of variance (ANOVA) followed by post hoc test]. (C)
Low-magnification (LM) and high-magnification confocal images of GFP
(green), MPC1 (red), and Ki67 (white) immunostaining in DG of
Nestin-GFP reporter mice. Scale bar, 10 μm. Arrowheads show MPC1
staining. (D) Quantification of MPC1 in Nestin-GFP–positive quiescent
and proliferative NSPCs. Data are shown as means ± SEM. **P < 0.01,
n = 3 mice (unpaired Student’s t test). (E) Experimental outline of
quiescence induction. (F) Relative mRNA expression of Mpc1 in
proliferative, quiescent, and ex-quiescent NSPCs. Bars represent
means ± SEM. **P < 0.01, n = 3 biological replicates (one-way ANOVA
followed by post hoc test). (G) Representative Western blots of MPC1
(12 kDa), MPC2 (14 kDa), and HSP70 (70KDa) in proliferative, quiescent,
and ex-quiescent NSPCs. MPC1 protein levels were normalized to
mitochondrial HSP70 levels and expressed as fold change to
proliferative NSPCs. Bars represent means ± SEM. ***P < 0.001, n = 3
biological replicates (one-way ANOVA followed by post hoc test). (H)
Experimental outline and images of pH3 and
4′,6-diamidino-2-phenylindole (DAPI) immunostaining in quiescent NSPCs
treated with UK5099. Scale bar, 50 μm. (I) Quantification of pH3^+ in
control or UK5099-treated quiescent NSPCs. Data represent means ± SEM.
*P < 0.05. n = 6 coverslips from three independent experiments (one-way
ANOVA followed by post hoc test).
Unexpectedly, when we analyzed scRNA-seq data comprising different
activity states of NSPCs in vivo ([59]15), we found that Mpc1
expression was highest in the more quiescent NSPC populations and
decreased with lineage progression ([60]Fig. 1B and fig. S1B),
suggesting that the import of pyruvate into mitochondria might play a
fundamental role in the regulation of NSPC quiescence. Staining for
MPC1 protein in Nestin-GFP reporter mice confirmed that MPC1 is indeed
highly expressed in radial glia–like quiescent NSPCs and reduced upon
proliferation ([61]Fig. 1, C and D).
To assess whether Mpc1 expression was dependent on the activity state
of NSPCs, we used cultured adult hippocampal NSPCs and applied an
established in vitro system, which mimics quiescence in a reversible
way ([62]Fig. 1E) ([63]18, [64]28, [65]29). Supporting the scRNA-seq
analysis, we found that the levels of Mpc1 mRNA and MPC1 protein were
significantly higher in quiescent NSPCs compared to
active/proliferating NSPCs ([66]Fig. 1, F and G). Mpc2/MPC2 followed
the same pattern ([67]Fig. 1G and fig. S1C). When the artificial
quiescence state of NSPCs was reverted back to a proliferative state,
Mpc1/MPC1 and MPC2 levels were significantly reduced to the levels of
proliferative NSPCs ([68]Fig. 1, F and G, and fig. S1C), indicating
that MPC levels are dynamically regulated with activity state. OXPHOS
complexes followed the same pattern, with increased protein levels in
quiescent NSPCs (fig. S1, D and E). Furthermore, morphometric analysis
of mitochondria revealed that quiescent NSPCs contained primarily fused
and elongated mitochondria, which are indicative of highly active
OXPHOS ([69]30, [70]31), whereas proliferative NSPCs had more
fragmented mitochondria (fig. S1, F and G). Together, these findings
show that quiescent NSPCs have high levels of MPC and an elongated
mitochondrial network.
Given the high MPC levels, we next tested whether inhibiting MPC would
affect the quiescence of these cells. We used the specific MPC
inhibitor UK5099 ([71]32) either during quiescence induction
([72]Fig. 1, H and I) or after established quiescence (fig. S1, H and
I). Notably, UK5099 significantly increased the number of cycling and
dividing Ki67 and phospho-histone 3 (pH3)–positive NSPCs, suggesting
that MPC activity is indeed required for maintenance of quiescence and
that its blockage leads to NSPC activation in vitro. To confirm these
findings, we isolated and expanded NSPCs from adult Mpc1 floxed mice
[Mpc1 fl/fl ([73]33)] and wild-type (WT) control mice. Infection with a
Cre-GFP retrovirus (RV) during proliferation led to a clear reduction
in MPC1 protein in the Mpc1 fl/fl NSPCs (hereafter called MPC1 cKO)
compared to virus-treated Mpc1 WT NSPCs (MPC1 WT), as shown by
immunohistochemistry (fig. S1J). Ablation of Mpc1 significantly
increased the number of proliferating NSPCs under quiescence
conditions, confirming the findings obtained with UK5099 (fig. S1, K
and L).
Mpc1 deletion in NSPCs in vivo leads to increased numbers of progeny and
triggers NSPC proliferation
We next addressed whether MPC is important for the maintenance of
quiescent NSPCs in vivo. To delete Mpc1 in adult NSPCs, we crossed Mpc1
fl/fl mice with a tamoxifen-inducible glial fibrillary acidic protein
(GFAP)–Cre recombinase line [hGFAP-CreERT2 ([74]34)]. To visualize the
recombined cells and their progeny, we additionally crossed these
inducible Mpc1 cKO mice with an inducible tdTomato lineage tracing line
[tdTom ([75]35)], resulting in triple transgenic mice hereafter called
MPC1 cKO-tdTom mice ([76]Fig. 2A). Control mice were generated the same
way but carried WT alleles of the Mpc1 gene (MPC1 WT-tdTom mice). Cre
expression was induced by administration of tamoxifen at postnatal day
30 (P30), leading to the deletion of Mpc1 in GFAP-expressing cells from
cKO-tdTom mice, including NSPCs ([77]Fig. 2A).
Fig. 2. Mpc1 deletion in NSPCs in vivo leads to increased numbers of progeny
and triggers NSPC proliferation.
[78]Fig. 2.
[79]Open in a new tab
(A) Generation of MPC1 cKO-tdTom and MPC1 WT-tdTom mice and
experimental timeline of tamoxifen (TAM) injections (green arrows) and
data analysis (black arrow). (B) LM confocal images of TAM-induced
td-Tom recombination (red), Ki67 (green), and DAPI (blue)
immunostaining at P38 in the DG. Scale bars, 10 μm. (C)
High-magnification (HM) confocal images of TAM-induced td-Tom
recombination (red), Ki67 (green), doublecortin (DCX; white), and DAPI
(blue) immunostaining at P38 in the DG. White arrowheads show the
colocalization of Ki67 and tdTom in a single confocal plane. Yellow
arrowheads show type 1 cell processes. Scale bars, 10 μm. (D)
Quantification of Ki67-positive (Ki67^+) tdTom^+ cells over total
number of tdTom^+ cells in the DG. Data are shown as means ± SEM.
***P < 0.001, n = 4 mice per group (unpaired Student’s t test). (E)
Quantification of type 1 Ki67^+ DCX- tdTom^+ cells over the total
number of tdTom^+ cells. Data are shown as means ± SEM. *P < 0.05,
n = 4 mice per group (unpaired Mann-Whitney t test). (F) Illustration
shows type 1 cell activation following Mpc1 deletion.
To determine the effect of Mpc1 deletion on the activation of NSPCs in
vivo, we analyzed the number of tdTomato-positive cells at P38
([80]Fig. 2A), a time point where MPC1 protein was clearly reduced in
MPC1 cKO-tdTom NSPCs (fig. S2B). While the total number of
tdTom-positive (tdTom^+) cells was not increased at P38, i.e., 8 days
after the first tamoxifen injections (fig. S2C), the proportion of
activated/proliferative tdTom^+ cells was significantly increased, as
revealed by the cell cycle marker protein Ki67 ([81]Fig. 2, B to D). As
several subtypes of NSPCs in the DG can proliferate, we categorized the
Ki67^+/tdTom^+ cells according to the previously described nomenclature
([82]36) into type 1 cells (radial glia–like cells) and type 2a cells
(no radial process, more proliferative progenitors). Type 2a cells were
further distinguished from committed neuroblasts by the absence of the
protein doublecortin (DCX), an immature neuronal marker (fig. S2A).
Whereas the number of Ki67^+ tdTom^+ type 2a cells was comparable
between control and MPC1 cKO-tdTom mice (fig. S2D), the number of
Ki67^+tdTom^+ cell type 1 cells was significantly increased in the KO
mice ([83]Fig. 2, C and E), indicating a substantial activation of
quiescent NSPCs upon Mpc1 deletion ([84]Fig. 1F).
Together, these data show that Mpc1 deletion in NSPCs leads to an
increased activation and proliferation of NSPCs, resulting in higher
numbers of progeny generated. These findings are in line with the
increased activation of quiescent NSPCs seen in vitro upon MPC blockage
and Mpc1 cKO ([85]Fig. 1, H and I, and fig. S1, K and L).
Increased lactate levels are not essential for the activation of quiescent
NSPCs
We next assessed the metabolic profile of quiescent NSPCs by using
Seahorse technology. A fuel flexibility test showed that the oxygen
consumption rate (OCR) in quiescent NSPCs depends mainly on glucose.
However, glutamine and fatty acids also contributed as oxidative fuel
sources (fig. S3A), suggesting that quiescent NSPCs are flexible in
their use of substrates for energy and biosynthetic metabolism. OCR was
markedly impaired after acute injection of UK5099 under basal
conditions in quiescent NSPCs ([86]Fig. 3A), while the inhibition of
FAO and glutaminolysis using etomoxir and
bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide (BPTES),
respectively, decreased OCR to a lesser extent (fig. S3B). These data
suggest that quiescent NSPCs use pyruvate as the main substrate for
mitochondrial respiration and that at least a part of the pyruvate
produced by glycolysis is transported into mitochondria.
Fig. 3. MPC inhibition in quiescent NSPCs leads to increased intracellular
aspartate levels.
[87]Fig. 3.
[88]Open in a new tab
(A) Experimental outline. Bottom: OCR and ECAR changes after acute MPC
inhibition with 1 μM UK5099. *P < 0.05, **P < 0.01, n = 3 biological
replicates; gray dots represent individual Seahorse wells (unpaired
Student’s t test). (B) Extracellular lactate in quiescent NSPCs. Data
represent means ± SEM, ***P < 0.001, n = 9 samples from three
independent experiments (unpaired Student’s t test). (C) Top:
Experimental outline. Bottom: Quantification of
5-ethynyl-2'-deoxyuridine-positive cells in control-, Ldh-i–,
Ldh-i + UK5099–, and UK5099-treated quiescent NSPCs. Data represent
means ± SEM *P < 0.05, ***P < 0.001. n = 9 coverslips from three
independent experiments (D) Top: Experimental outline. Bottom:
Principal components analysis (PCA) of untargeted metabolomics from
control- and UK5099-treated quiescent NSPCs. Gray, control-treated
cells; red, UK5099-treated cells. n = 4 biological replicates. (E)
Relative intensity of selected metabolites in control- and
UK5099-treated quiescent NSPCs. Bars represent means ± SEM normalized
to control condition. *P < 0.05, n = 4 biological replicates. Cit,
citrate; Aco, aconitate; aKG, alpha-ketoglutarate; Suc, succinate; Fum,
fumarate; Mal, malate; Gln, glutamine; Glu, glutamate; Asp, aspartate;
Asn, asparagine; and Ala, alanine. (F) Main metabolic pathways involved
in the synthesis and utilization of aspartate. OAA, oxaloacetate. (G)
Relative mRNA expression of Asns, Got1, Got2, and Slc1a3 in control-
and UK5099-treated NSPCs; bars represent means ± SEM. *P < 0.05, N = 3
independent experiments. Statistics were computed on ΔCT values
(unpaired Student’s t test). (H) Top: Experimental outline. Bottom:
Images of pH3 and DAPI immunostaining in quiescent NSPCs treated with
UK5099, aspartate and UK5099 + aspartate. Scale bar, 50 μm. (I)
Quantification of pH3-positive cells in control-, aspartate-, UK5099-
and aspartate + UK5099–treated quiescent NSPCs. Data represent
means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005. n = 6 coverslips from
two independent experiments (one-way ANOVA followed by post hoc test).
Acute inhibition with UK5099 also resulted in a significant increase in
the extracellular acidification rate (ECAR; [89]Fig. 3A), whereas
blocking FAO and glutaminolysis had no effect on ECAR (fig. S3C). These
results suggest increased lactate secretion upon inhibition of MPC,
which we confirm by direct measurement of lactate in the medium of
quiescent NSPCs treated with UK5099 ([90]Fig. 3B). Previous studies
have shown that exogenous lactate can promote NSPC proliferation
([91]37–[92]39). We therefore speculated that higher extracellular
lactate levels might be the cause of increased proliferation after MPC
inhibition. We therefore tested whether addition of lactate during
quiescence induction could mimic the effect of UK5099. However, the
proliferation rate of quiescent NSPCs did not change with the addition
of lactate (fig. S3D), although the same dose of lactate was sufficient
to increase the proliferation of NSPCs under proliferating conditions
(fig. S3E). To demonstrate that an increased lactate concentration per
se is not sufficient to explain the effects of UK5099 on quiescent
NSPCs, we next used an established lactate dehydrogenase inhibitor [GSK
2837808A ([93]40), here called Ldh-i], which prevents the conversion of
pyruvate to lactate. The lactate concentration in the medium of
quiescent NSPCs treated with both Ldh-i and UK5099 was significantly
lower than in cells treated with UK5099 alone (fig. S3F), indicating
that Ldh-i reduced lactate production as intended. If the conversion of
pyruvate to lactate were required for the increased proliferation after
MPC inhibition, then concomitant treatment with Ldh-i should at least
partially prevent the UK5099 effect. However, we found no difference in
the proliferation of NSPCs under quiescence condition when treated with
UK5099 alone or with UK5099 and Ldh-i simultaneously ([94]Fig. 3C).
Together, these data suggest that elevated lactate levels are not
essential for the activation of quiescent NSPCs.
MPC inhibition in quiescent NSPCs leads to increased intracellular aspartate
levels
To analyze the metabolic changes resulting from inhibition of
mitochondrial pyruvate import in quiescent NSPCs, we performed
untargeted metabolomics on NSPCs cultured in the presence or absence of
UK5099 during the induction of quiescence ([95]Fig. 3D). Principal
components analyses revealed that samples clustered according to the
treatment, suggesting that MPC inhibition drives profound metabolic
changes in quiescent NSPCs ([96]Fig. 3D and fig. S3G). Pathway
enrichment analysis showed that UK5099 significantly affected TCA cycle
and amino acid metabolism (fig. S3H). While intermediates of glycolysis
were not significantly changed (fig. S3I), TCA intermediates, in
particular, citrate and succinate levels, were clearly reduced compared
to untreated cells ([97]Fig. 3E). Notably, malate and fumarate were not
significantly different. Furthermore, MPC inhibition led to a marked
increase in intracellular aspartate ([98]Fig. 3E). Overall, these
measurements indicate an impairment in TCA activity and a rewiring of
amino acid metabolism leading to high aspartate levels.
The intracellular pool of aspartate is maintained by the activity of
two transaminases, the cytosolic enzyme glutamic-oxaloacetic
transaminase 1 (GOT1) and the mitochondrial GOT2. GOT1 uses aspartate
to generate glutamate and oxaloacetate, while GOT2 uses these two
substates to synthesize α-ketoglutarate and aspartate ([99]Fig. 3F). We
thus measured mRNA expression of the two transaminases. Got1 was
significantly increased in quiescent NSPCs treated with UK5099
([100]Fig. 3G), whereas Got2 was only slightly increased. In addition,
asparagine synthase (Asns), which converts aspartate to asparagine, was
also increased ([101]Fig. 3G). These data suggest that increased
expression of Got transaminases upon MPC inhibition contributes to the
increased aspartate levels.
Cells can also uptake aspartate through the glutamate-aspartate plasma
membrane transporter solute carrier family 1 member 3 (SLC1A3, also
known as GLAST) ([102]Fig. 3F). Quiescent NSPCs express already high
levels of this transporter ([103]41), and no further increase occurred
upon UK5099 treatment ([104]Fig. 3G). Exogenous aspartate or
overexpression of SLC1A3 has been shown to be sufficient to support
proliferation of cells lacking electron transport chain (ETC) activity
([105]42, [106]43). To test whether the measured increase in aspartate
upon MPC inhibition could be responsible for the activation of
quiescent NSPCs, we treated NSPCs with different concentrations of
aspartate during quiescence induction and observed a significant
increase in the number of proliferating, pH3-positive cells
([107]Fig. 3, H and I, and fig. S3J). Moreover, simultaneous treatment
with UK5099 and 5 mM aspartate further increased the ratio of
pH3-positive NSPCs ([108]Fig. 3, H and I). Beyond its role as an amino
acid in proteins, aspartate is also required for nucleotide synthesis
([109]44). We found that some of the nucleotide synthesis enzymes are
indeed up-regulated with UK5099 treatment (fig. S3K), indicating that
increased aspartate levels might be used to promote nucleotide
synthesis. Together, these data show that inhibition of MPC causes an
increase in aspartate levels in quiescent NSPCs, which could result
from increased GOT activity and/or increased import. This increase in
aspartate is sufficient to drive NSPC activation.
MPC-deficient NSPCs can generate neurons through a shift in their metabolism
Once activated, NSPCs can proliferate and eventually differentiate into
neurons. We therefore next tested the effect of UK5099 in already
active/proliferative NSPCs. Similar to quiescent NSPCs, active NSPCs
showed decreased OCR and increased ECAR upon treatment with UK5099
([110]Fig. 4A), suggesting that these cells also use pyruvate to drive
OXPHOS despite lower levels of MPC1 and MPC2 than quiescent NSPCs
([111]Fig. 1, F and G). Untargeted metabolomics showed a metabolic
shift ([112]Fig. 4B), with a significant decrease of TCA intermediates
([113]Fig. 4C) and a shift toward increased production of aspartate and
asparagine with UK5099 treatment ([114]Fig. 4C), while no changes in
the intermediates of glycolysis were found (fig. S4A). However, despite
these changes in mitochondrial metabolism, UK5099 did not alter the
proliferation rate of already active NSPCs ([115]Fig. 4D).
Fig. 4. MPC-deficient NSPCs can generate neurons through a shift in their
metabolism.
[116]Fig. 4.
[117]Open in a new tab
(A) Experimental outline. Bottom: OCR and ECAR measurements in control
and 1 μM UK5099-treated NSPCs. Bars represent means ± SEM. **P < 0.01.
n = 3 biological replicates (unpaired Student’s t test), gray dots
represent individual wells. (B) Top: Experimental outlines. Bottom: PCA
of untargeted metabolomics from control- and UK5099-treated
proliferative NSPCs. n = 4 biological replicates. (C) Relative
intensity of selected metabolites in control- and UK5099-treated cells.
Bars represent means ± SEM. * adjusted P value <0. 1, n = 3 biological
replicates. (D) Quantification of pH3^+ cells in control- and
UK5099-treated proliferative NSPCs. Data represent means ± SEM. Not
significant (ns), n = 6 coverslips from two independent experiments.
(unpaired Student’s t test). (E) Illustration shows the MPC1 cKO system
in cultured NSPCs. (F) LM and HM confocal images of MPC1 (green) and
DAPI (blue) immunostaining in MPC1 WT and MPC1 cKO NSPCs. Scale bar,
LM, 20 μm; HM, 5 μm. (G) Quantification of MPC1 intensity in MPC1 WT
and MPC1 cKO NSPCs. *P < 0.05 n = 3 electroporations, data show
means ± SEM, unpaired Student’s t test. (H) Quantification of pH3^+
cells in MPC1 WT and MPC1 cKO NSPCs. Data represent means ± SEM. n = 3
electroporations. (I) LM and HM confocal images of neurons (MAP2ab,
green) and DAPI (blue). Scale bar, LM, 20 μm; HM, 5 μm. (J)
Quantification of MAP2ab-positive cells in MPC1 WT and MPC1 cKO cells.
*P < 0.05, n = 3 electroporations, means ± SEM (unpaired Student’s t
test). (K) Quantification of dendritic length of MAP2ab-positive
neurons. n = 13 neurons per group (unpaired Student’s t test). (L) GO
term analysis of the up-regulated genes in MPC1-cKO NSPCs.
Neuronal differentiation is associated with an increased expression of
genes involved in OXPHOS ([118]14, [119]15). As glucose- or
lactate-derived pyruvate is thought to be the major fuel for OXPHOS
during differentiation, MPC-mediated pyruvate import should thus also
play an important role during NSPC differentiation. We therefore tested
whether active NSPCs could differentiate into neurons in the presence
of UK5099. We found a clear reduction in the total number of cells
after 7 days of differentiation in the presence of UK5099 (fig. S4B),
indicating that mitochondrial pyruvate might regulate cell survival or
cell cycle exit during the differentiation process. However,
unexpectedly, UK5099-treated NSPCs were still able to give rise to
microtubule-associated protein 2ab (MAP2AB)–positive neurons in a
similar ratio to total cell numbers as untreated NSPCs (fig. S4, C and
D). These data highlight a large metabolic flexibility of NSPCs, which
seem to use other metabolic pathways, in particular, amino acid
metabolism, to sustain their energy needs for proliferation and
differentiation when pyruvate import into mitochondria is disrupted.
To confirm these important findings, we used NSPCs from Mpc1 fl/fl mice
and WT control mice and induced Cre-mediated recombination through
electroporation of an mCherry-Cre–expressing plasmid ([120]Fig. 4E) and
obtained purely transfected cell populations by fluorescence-activated
cell sorting (FACS; fig. S4E). Three days after Cre-mediated deletion
of Mpc1, MPC1 protein levels were strongly reduced in proliferating
MPC1 cKO NSPCs compared to MPC1 WT NSPCs ([121]Fig. 4, F and G).
Similar to UK5099-treated active NSPCs, Mpc1 deletion did not affect
proliferation of activate NSPCs ([122]Fig. 4H). We next assessed
whether MPC1 cKO NSPCs were able to differentiate into neurons. While
we also found a similar reduction in total cell numbers as in
UK5099-treated NSPCs after 7 days of differentiation (fig. S4F), the
deletion of Mpc1 in active NSPCs did not affect the differentiation and
the maturation into MAP2AB-positive neurons. MPC1-cKO NSPCs generated
even slightly more MAP2AB-positive neurons per total number of cells
compared to control NSPCs ([123]Fig. 4, I to K).
To get further insights into the increased neuron production, we next
performed RNA-seq on active MPC1 WT and MPC1 cKO NSPCs. Commonly used
NSPC markers were comparable between the two groups (fig. S4, G to I),
indicating that MPC1 cKO cells retained the main features of NSPCs
despite the lack of MPC1. Among the differentially expressed genes, 160
genes were up-regulated, and 34, including Mpc1, were down-regulated in
the MPC1 cKO NSPCs (fig. S4J). Gene ontology (GO) analysis of the
up-regulated genes showed a clear enrichment of GO terms involved in
neuronal differentiation ([124]Fig. 4L), while the down-regulated genes
did not enrich in a specific GO category. Together, these results
indicate that blocking the import of pyruvate into the mitochondria for
OXPHOS does not affect neuronal differentiation of NSPCs, contrary to
what would have been predicted.
Mpc1 deletion in NSPCs increases neurogenesis in vivo
We next tested whether deletion of Mpc1 in adult NSPCs in vivo would
affect their differentiation capacity 30 days after the first tamoxifen
injection ([125]Fig. 5B). Cre-mediated Mpc1 deletion at P30 led to a
significant reduction in Mpc1 mRNA and MPC1 protein in the hippocampus,
as shown by quantitative reverse transcription polymerase chain
reaction (qRT-PCR) and Western blot on whole hippocampal tissue of MPC1
cKO-tdTom and control mice at P60 (fig. S5, A to C). MPC2 was also
down-regulated (fig. S5C), as has been previously shown in other MPC1
KO cell types ([126]11, [127]45–[128]48).
Fig. 5. Mpc1 deletion in NSPCs increases neurogenesis in vivo.
[129]Fig. 5.
[130]Open in a new tab
(A) LM and HM confocal images of TAM-induced tdTom recombination
(tdTom, red), NeuN (green), and DAPI (blue) immunostaining. Scale bars,
10 μm. (B) Experimental scheme. (C) Quantification of tdTom^+ cells per
DG sections. Data are shown as means ± SEM. ***P < 0.001, n = 4 mice
per group (unpaired Student’s t test). (D) Quantification of
DCX + tdTom^+ cells per DG sections. Data show means ± SEM. *P < 0.05,
n = 4 mice per group (unpaired Student’s t test). (E) Quantification of
NeuN+tdTom^+ cells. Data show means ± SEM. ***P < 0.001, n = 4 mice per
group (unpaired Student’s t test). (F) Images of in vivo injections of
RV Cre-GFP in the DG. Scale bars, 250 μm. (G) Confocal images of
Cre-induced td-Tom + newborn neurons. Arrowheads indicate the
colocalization between nuclear GFP and tdTom signal. Scale bars, 10 μm.
Confocal image of spines in newborn neurons. Scale bars, 1 μm. (H) Top:
Experimental scheme. Bottom: Sholl analysis of dendritic complexity of
newborn neurons. n = 4 mice per group. (I) Quantification of the
dendritic length of newborn neurons. n = 4 mice per group (unpaired
Student’s t test). (J) Spine density of newborn neurons. n = 4 mice per
group. (K) LM and HM confocal images and a single plane image of
TAM-induced td-Tom recombination (red), DCX (white), NeuN (green), and
DAPI (blue) immunostaining. Scale bars, 10 μm. (L) Top: Experimental
scheme. TAM injection (green arrow) at 9 to 11 months (“middle-aged
mice”), and data analysis after 1 month (black arrow). Bottom:
Quantification of NeuN+dTom^+ cells per DG sections. Data show
means ± SEM. *P < 0.05, n = 3 to 4 mice per group (unpaired Student’s t
test).
Immunohistological analyses showed that the overall numbers of tdTom^+
cells in the DG were significantly increased in MPC1 cKO-tdTom mice
compared to control mice 1 month after Mpc1 deletion ([131]Fig. 5, A
and C). This increase was due to increased numbers of neuroblasts and
immature neurons marked by DCX^+/tdTom^+ ([132]Fig. 5D and fig. S5D),
as well as increased numbers of mature newborn neurons (NeuN^+/tdTom^+)
([133]Fig. 5, A and E). These data show that the generation of newborn
neurons in vivo is not impaired by Mpc1 deletion in NSPCs. On the
contrary, the lack of MPC1 in NSPCs even led to a significant increase
in the generation of newborn neurons.
To address whether the increased neurogenesis was due to a
cell-autonomous Mpc1 cKO effect in already activated NSPCs, we
performed stereotaxic injection of an RV encoding a Cre-recombinase
fused to a GFP fluorophore into the DG of mice carrying floxed or WT
Mpc1 alleles crossed to the tdTom lineage tracing line ([134]Fig. 5F).
As previously reported, this viral Cre-delivery strategy targets only
activated NSPCs ([135]49). This approach results in MPC1-deficient
NSPCs (MPC1 cKO-GFP) or control NSPCs (MPC1 WT-GFP) in an otherwise
unaltered microenvironment, and all progeny are labeled with tdTom and
GFP. One month after virus injection, mature neurons were found in both
MPC1 cKO-GFP and MPC1 WT-GFP mice ([136]Fig. 5G). Despite the lack of
MPC1, the neurons in the MPC1 cKO-GFP mice had similar morphology to
the neurons in MPC1 WT-GFP mice, as assessed by Scholl analysis and
dendritic length measurements ([137]Fig. 5, H and I). Similarly, total
spine numbers were not changed ([138]Fig. 5J), and the maturity of the
spines was normal (fig. S5E). These astonishing findings show that even
with disrupted oxidation of glucose-derived pyruvate, NSPCs can give
rise to normal mature newborn neurons. This suggests that NSPCs and
their progeny display metabolic flexibility and can compensate for
impaired pyruvate import.
Mpc1 deletion also increases neurogenesis in advanced age mice
Hippocampal neurogenesis drastically drops with age ([139]50), with
more than fourfold reduction of cycling NSPCs at 9 months compared to 2
months of age ([140]51). Several recent studies suggest that this is
due to increased dormancy of NSPCs ([141]52–[142]54). When querying
scRNA-seq data from a recent publication that compared dormant,
resting, and proliferating NSPCs in vivo ([143]55), we found that Mpc1
and Mpc2 were most highly expressed in dormant cells and lowest in
proliferating NSPCs (fig. S5F). As Mpc1 deletion leads to an activation
of quiescent NSPCs in young adult mice ([144]Fig. 2, C, E, and F), we
next tested whether deletion of Mpc1 would have similar effects on
dormant NSPCs from older animals. Mpc1 deletion was induced in 9- to
11-month-old MPC1 cKO-tdTom mice, and the number of newborn neurons was
assessed 30 days later. The number of tdTom^+ neurons was significantly
increased in the aged MPC1 cKO tdTom mice compared to the MPC1 WT tdTom
mice ([145]Fig. 5, K and L), suggesting that in mice with advanced
aged, NSPCs are also activated upon Mpc1 ablation, leading to an
increase in neurogenesis.
DISCUSSION
One of the most important regulatory steps of adult neurogenesis is the
decision of NSPCs to maintain quiescence or enter an active state. This
decision results from intrinsic and extrinsic instructions, which are
still not fully understood ([146]2). We here show that the maintenance
of quiescence in adult NSPCs is dependent on mitochondrial pyruvate
metabolism and requires MPC activity. MPC, which imports pyruvate into
the mitochondria, is highly expressed in adult NSPCs, and its loss of
function leads to an activation of quiescent NSPCs and increased
neurogenesis (summarized in [147]Fig. 6). Recent studies have shown
increased proliferation upon Mpc1 deletion in other stem cells, such as
intestinal and hair follicle stem cells ([148]10, [149]11).
Furthermore, tumor cells slowed down proliferation after overexpression
of Mpc1 ([150]48, [151]56), suggesting that there might be a common
regulation of cellular activity by MPC.
Fig. 6. Graphical summary of the presented data.
[152]Fig. 6.
[153]Open in a new tab
Given the general idea that stem cells are mostly glycolytic, whereas
their differentiated progeny use OXPHOS, the involvement of MPC in the
quiescence of NSPCs shown here is unexpected as this carrier is
generally considered a marker of oxidative metabolism. Our results
suggest that quiescent type 1 cells display a more oxidative metabolism
than commonly thought, and the import of pyruvate into the mitochondria
is necessary to maintain their quiescent state. Accordingly, we and
others have reported that these cells have an elaborate mitochondrial
network and can use fatty acids to fuel OXPHOS ([154]14,
[155]18–[156]21, [157]57). A possible explanation for these
discrepancies comes from a recent study comparing proteomic and
transcriptomic levels of several metabolic proteins: Wani and
colleagues ([158]19) could demonstrate that changes in TCA, FAO, and
OXPHOS proteins were poorly reflected at the transcriptional levels,
while glycolytic enzymes showed better correlation between protein and
mRNA expression. Our findings on the importance of MPC for quiescent
NSPCs opens different perspectives to better understand the maintenance
of quiescence from a metabolic point of view.
The detailed molecular mechanisms by which mitochondrial pyruvate
controls type 1 stem cell proliferation remain to be elucidated.
Although lactate levels clearly change upon MPC inhibition, our
experiments using lactate supplementation and inhibition of lactate
dehydrogenase suggest that increased concentration of lactate are not
essential for the activation of NSPCs. Lactate production triggers the
activation of hair follicle stem cells ([159]10), indicating that the
effect driven by lactate on the activation and proliferation depends on
the stem cell types. The inhibition of pyruvate transport into
mitochondria has been shown to induce a metabolic shift toward
oxidation of other substrates, in particular, amino acids, which in
turn rewire the metabolism of the cells ([160]46, [161]58, [162]59). We
also observe major metabolic changes when blocking mitochondrial
pyruvate import, which led to the activation of quiescent NSPCs. We
show that inhibition of MPC induces an increase of intracellular
aspartate despite a significant decrease in OCR and TCA metabolites.
Aspartate can have different sources, thus a metabolic flux analysis
would be required to elucidate the detailed mechanism by which NSPCs
increase their intracellular aspartate levels when treated with UK5099.
Using metabolic flux analyses, other studies have shown that MPC
inhibition in different cell types leads to an increase of aspartate
synthesis through mitochondrial oxidation of glutamine or glutamate
([163]46, [164]58, [165]59). Whether this is also the case in NSPCs
remains to be determined. Intriguingly, we also found that MPC
inhibition induces a significant increase of Got1 expression and a
slight increase of Got2, two key components of the malate-aspartate
shuttle system. These data suggest that quiescent NSPCs might
up-regulate the malate-aspartate shuttle upon MPC inhibition to
transfer cytosolic NADH [reduced form of nicotinamide adenine
dinucleotide (NAD^+)] into mitochondria for energy production and to
regenerate NAD^+ to balance the cellular NAD^+/NADH pool. Since it has
been shown that in the context of a decline in ETC activity, the
recycling of NAD^+ is necessary for cell proliferation ([166]60), the
malate aspartate shuttle activation might also be important in cells
lacking MPC function. ETC deficiency can also lead to a reversal of
GOT1 activity, leading to GOT1-dependent reductive aspartate synthesis,
as, for instance, shown by two recent studies ([167]42, [168]43). In
quiescent NSPC lacking MPC activity, the ETC is expected to be
functional. However, we cannot exclude that GOT1 activity could also be
reversed because of fewer electrons feeding complex I of ETC and
decreased OXPHOS.
Nevertheless, aspartate per se is a key amino acid for cell
proliferation not only for protein synthesis but also for purine
nucleotide and pyrimidine nucleotide synthesis ([169]43). Exogenous
aspartate is sufficient to sustain proliferation even in the absence of
a functioning ETC ([170]43) and is also a limiting metabolite for
cancer cell proliferation ([171]61). Furthermore, aspartate levels have
been shown to directly influence hematopoietic stem cell proliferation
([172]44). Our findings that the addition of exogenous aspartate mimics
the effect of UK5099 on quiescence exit of NSPCs are thus in line with
these previous findings. The additive effect of aspartate and UK5099
suggest that both mechanisms, aspartate uptake and/or aspartate
biosynthesis, contribute to the intracellular pool of aspartate.
Despite its clear influence on NSPC activation, it remains to be
clarified how exactly increased aspartate is exerting its effect. A
possible explanation is that aspartate is used for the synthesis of
purine/pyrimidine and asparagine, similar to findings in hematopoietic
stem cells ([173]44). Our metabolomic and mRNA expression data point in
this direction, showing an increase of asparagine synthetase and
asparagine synthesis as well as slight increase in nucleotide synthesis
enzymes in quiescent NSPCs upon MPC inhibition. The additive effect of
aspartate and UK5099 could also indicate that besides increasing
intracellular aspartate, MPC inhibition might have additional effects
on the activation of quiescent NSPCs.
A shift in the balance of NSPC quiescence and activation is expected to
affect net neurogenesis as the production of new neurons is the default
fate of activated adult hippocampal NSPCs ([174]1). Accordingly, we
found a substantial increase in the number of newborn neurons upon Mpc1
deletion. However, given the commonly accepted view that most neurons
heavily rely on glucose- and lactate-derived pyruvate for OXPHOS
([175]62, [176]63), such an increase in morphologically normal neurons
is astonishing, as one would expect that newborn neurons cannot mature
properly without pyruvate oxidation. These results suggest that newborn
neurons might have a higher metabolic flexibility than previously
thought if their preferential metabolic pathway is blocked. In line
with this, it has been shown that several types of mature neurons are
indeed able to sustain their functions by shifting their metabolism to
other substrates, which can fuel OXPHOS when MPC is inhibited ([177]59,
[178]64, [179]65). We show that MPC1 cKO NSPCs are able to generate
mature neurons in vitro and in vivo. These findings highlight a new
form of metabolic plasticity used by NSPCs to differentiate into mature
neurons. MPC1 cKO further seems to prime NSPCs toward neuronal
differentiation, as our RNA-seq data suggest. These data illustrate
that the metabolic state might directly influence fate decision.
Whether this neuronal priming is achieved via epigenetic modifications
in MPC1 cKO NSPCs remains to be determined.
Hippocampal neurogenesis decreases strongly with age because of
terminal differentiation and increased deep quiescence of NSPCs
([180]52, [181]53). Our data show that MPC deletion is also sufficient
to wake up these more dormant NSPCs in aged mice, resulting in
increased number of newborn neurons. Supporting these findings, we
found that Mpc1 is highest in dormant NSPCs in a recent scRNA-seq study
comparing different states ([182]55), which is also in line with our in
vitro quiescence data.
In conclusion, we have shown that quiescence of NSPCs is an actively
maintained state, requiring MPC-mediated import of pyruvate into the
mitochondria. MPC inactivation leads to activation of quiescent NSPCs
and subsequently increased neurogenesis, even in aged mice. Our
findings thus describe a mechanism through which mitochondrial
metabolism controls NSPC function.
MATERIALS AND METHODS
Animals
All studies were approved by the Service de la consommation et des
affaires vétérinaires of Geneva in Switzerland. Mice were group-housed
with littermates in standard housing on a 12:12-hour light/dark cycle.
Nestin-GFP reporter mice ([183]66) have been obtained from
S. Jessberger (University of Zürich), hGFAP-CreERT2 mice ([184]34) from
N. Toni (University Hospital Lausanne, CHUV), Mpc1 fl/fl mice ([185]33)
from E. Taylor (University of Iowa), and tdTom fl-STOP-fl [Ai14
([186]35)] from I. Rodriguey (University of Geneva). All mice were on a
C57BL/6 background. Genotyping was performed on DNA extracted from
phalange biopsies using the following primers: hGFAP-CreERT2: F-CAG GTT
GGA GAG GAG ACG CAT CA, R-CGT TGC ATC GAC CGG TAA TGC AGG C; MPC1
fl/fl: F1-CCT ATT CTC TAG AAA GTA TAG GAA CTT CGT CGA, F2-GTG AGC CCA
GAG CTA CGA AGG ATC GGC, F3-GGA AAG AAA AAG GTG TCC AAT TTT AGC TCT
GCA; tdTom fl STOP-fl: F 5′-CTG TTC CTG TAC GGC ATG G-3′, R 5′-GGC ATT
AAA GCA GCG TAT CC-3′, tdTom WT/WT:-F 5′-AAG GGA GCT GCA GTG GAG TA-3′,
R 5′-CCG AAA ATC TGT GGG AAG TC-3′.
In vivo treatments
Tamoxifen (100 mg/kg; Sigma-Aldrich) was injected intraperitoneally
(ip) for five consecutive days (P30 to P35) to induce recombination in
vivo. The tamoxifen was dissolved in sunflower oil (S5007,
Sigma-Aldrich).
Analysis of transcriptomic resources
We queried available scRNA-seq data for Mpc1 expression ([187]15,
[188]27, [189]55). For the expression of Mpc1 in NSPCs, the original
unsupervised clustering and pseudotime subgroups (S1 to S5) from Shin
et al. were further divided in three subgroups: S1-S2, S3, and S4-S5,
reflecting, respectively, a more quiescent to proliferative state.
Virus preparation
CAG-Cre-GFP virus was produced as previously described ([190]67).
Briefly human embryonic kidney 293 T cells were transfected with a
pCAG-GFPcre, pCMV-gp, and pCMV-vsvg using Lipofectamine 2000 (no.
10696153, Thermo Fisher Scientific) in Opti-MEM (no.11520386, Thermo
Fisher Scientific). Forty-eight hours after transfection, the virus was
collected by filtering the cell culture medium through a 0.22-μm
filter. The filtrate was then concentrated twice using
ultracentrifugation at 19,400 rpm. The viral pellet was resuspended in
phosphate-buffered saline (PBS) and used for in vitro experiment and
stereotaxic injections.
Cell culture
Adult mouse NSPCs from hippocampus of 7-week-old C57/Bl6 (MPC1 WT) or
Mpc1 fl/fl (MPC1 cKO) male mice were isolated as previously described
([191]68). For propagation, cells were cultured as spheres in
Dulbecco’s modified Eagle medium (DMEM)/F12 GlutaMAX (no. 31331-028,
Invitrogen) supplemented with N2 (no. 17502048, Invitrogen), human
epidermal growth factor (EGF, 20 ng/ml; no. AF-100-15, Peprotech),
human fibroblast growth factor (FGF)–basic (20 ng/ml; no. 100-18B,
Peprotech), and heparin (5 μg/ml; no. H3149-50KU, Sigma-Aldrich).
Medium contained as well an antibiotic-antimycotic (no. 15240062,
Invitrogen). Medium was changed every second day. For proliferation
experiments, cells were plated (42,000 cells/cm^2) on glass coverslip
(no. 10337423, Thermo Fisher Scientific) coated with poly-l-ornithine
(50 μg/ml; no. [192]P36655, Sigma-Aldrich) and laminin (5 μg/ml; no.
L2020, Sigma-Aldrich) and grown as described above.
Quiescence was induced as previously described ([193]18). Proliferating
NSPCs (40,000 cells/cm^2) were plated on coated glass coverslip or
coated plastic cell culture well plate. Twenty-four hours later, medium
was changed to quiescence medium based on DMEM/F12 GlutaMAX,
supplemented with N2, human FGF-basic (20 ng/ml), heparin (5 μg/ml),
and bone morphogenetic protein 4 (50 ng/ml; no. 5020BP, R&D Systems).
Seventy-two hours after induction, cells were fully quiescent and used
for subsequent experiments. For UK5099 experiment, quiescence was fully
established for 3 days, and then cells were washed off, trypsinized,
and plated (220,000 cells/cm^2) on coated coverslips in fresh
quiescence medium.
For comparison, proliferative NSPCs were plated at a lower density
(13,000 cells/cm^2) under proliferative conditions. The quiescence
state was reversed by collecting quiescent NSPCs and replating them in
proliferative conditions (50,000 cells/cm^2) for 3 days ([194]Fig. 2A).
5-ethynyl-2′-deoxyuridine (EdU) pulse was performed by incubating
quiescent NSPCs with 10 μM EdU for 1 hour at 37°C before fixation.
For differentiation experiments, cells were plated on glass coverslip
(no. 10337423, Thermo Fisher Scientific) or cell culture plates coated
with poly-l-ornithine (50 μg/ml; no. [195]P36655, Sigma-Aldrich) and
laminin (5 μg/ml; no. L2020, Sigma-Aldrich) and cultured with a medium
containing one-fifth of human EGF and human FGF-basic, i.e., 2 ng/ml.
After 3 days, medium was change to medium without any human EGF and
human FGF-basic. Cells were fixed for 15 to 20 min with 4%
paraformaldehyde (PFA) after 6 to 7 days of differentiation. Cells were
washed twice for 10 min with PBS and stored at 4°C before staining.
Treatment with drugs and aspartate
Proliferative NSPCs were treated with either 1 or 10 μM UK5099 (no.
PZ0160, Sigma-Aldrich) for 72 hours. For quiescence experiments, NSPCs
were treated with 1 or 10 μM UK5099 (no. PZ0160, Sigma-Aldrich), 1 μM
lactate dehydrogenase A inhibitor (GSK 2837808A, Tocris, 5189/10), and
1, 2.5, or 5 mM aspartate (no. 1690.2, Roth) simultaneously with
quiescence induction. For exit experiments, quiescent NSPCs were
treated with UK5099 after establishment of quiescence.
For differentiation experiments, proliferative NSPCs were treated with
1 or 10 μM UK5099 for 48 hours. Then, cells were washed off and plated
(65,000 cells/cm^2) on coated coverslips in media containing one-fifth
of the growth factors and indicated UK5099 concentration. Similar to
UK5099 treatment, 1, 2.5, or 5 mM aspartate (no. 1690.2, Roth) was used
during the induction of quiescence.
Virus infection in vitro
Proliferative NSPCs were plated (58,000 cells/cm^2) on a cell culture
plate coated with poly-l-ornithine (50 μg/ml; no. [196]P36655,
Sigma-Aldrich) and laminin (5 μg/ml; no. L2020, Sigma-Aldrich). One day
after seeding, 0.5 μl of RV Cre-GFP virus was added to cells. Two days
after virus infection, cells (40,000 cells/cm^2) were replated on a
glass coverslip in a proliferative medium. Twenty-four hours after,
medium was changed to a quiescent medium (see the "Cell culture"
section). After 3 days of quiescence, cells were fixed using 4% PFA.
Extracellular lactate measurement
After 3 days of quiescence, medium was collected, spun down at 300g for
5 min, and snap-frozen before measurements. Lactate concentration was
assessed using a buffer containing 0.33 M glycine-semicarbazide
(Sigma-Aldrich, S2201), 0.015 M NAD (Roche, 12127973001), and lactate
dehydrogenase (70 U/ml; Roche, 10127876001). Lactate concentration was
assessed in an indirect way—by measuring the conversion of NAD to NADH
measured at 340 nm.
Metabolic fuel flux assays
OCR and ECAR were measured with the Seahorse XFe96 Analyzer (Agilent).
Four thousand NSPCs per well were plated on 96-well Seahorse XF96 cell
culture microplates with proliferative medium. After 24 hours,
quiescence was induced (see the “Cell culture” section). After three
full days of quiescence condition, cells were carefully washed with
unbuffered DMEM (no. 103575-100, Agilent) supplemented with 10 mM
glucose (no. 103577-100, Agilent), 2 mM glutamine (no. 103579-100,
Agilent), and 1 mM pyruvate (no. 103578-100, Agilent), and a final
volume of 180 μl of assay media was added to cells. The plate was
allowed to equilibrate in a non-CO[2] incubator at 37°C for 1 hour
before analysis.
Mito Fuel Flex Test (no. 10360-100, Agilent) was performed according to
manufacturer’s protocol. Baseline OCR was monitored for 14 min followed
by different combinations of inhibitor injections as reported in
[197]Table 1 with OCR measurements. The Mito Fuel Flex Test inhibits
the import of the three main metabolic substrates—pyruvate, glutamine,
and fatty acids—using UK5099 (2 μM; no. PZ0160, Sigma-Aldrich), BPTES
(3 μM; no. SML0601, Sigma-Aldrich), and etomoxir (4 μM; no. E1905,
Sigma-Aldrich), respectively.
Table 1. Combination of inhibitors.
Metabolic test Injection no. 1 Injection no. 2
Pyruvate dependency UK5099 Etomoxir + BPTES
Pyruvate capacity Etomoxir + BPTES UK5099
Glutamine dependency BPTES UK5099 + etomoxir
Glutamine capacity UK5099 + etomoxir BPTES
Fatty acid dependency Etomoxir UK5099 + BPTES
Fatty acid capacity UK5099 + BPTES Etomoxir
[198]Open in a new tab
Dependency was calculated as
[MATH: Dependency(%)=(Baseline
OCR−target inhibitor
OCRBaseline OCR−All inhibitors
OCR)∗100 :MATH]
Capacity was calculated as
[MATH: Capacity(%)=[1−(Baseline OCR−other 2 inhibitors
OCRBaseline OCR−All inhibitors
OCR)]∗100 :MATH]
Flexibility was calculated as
[MATH: Flexibility(%)=%Capacity−%dependency :MATH]
To assess the effect of MPC1 inhibition, baseline OCR and ECAR were
measured after 14 min in quiescent NSPCs. Different combinations of
drugs—UK5099 (1 μM), etomoxir (20 μM), BPTES 3 (μM), etomoxir/BPTES,
UK5099/etomoxir, or UK5099/BPTES—were injected, and OCR and ECAR were
measured over 40 min. The change in OCR or ECAR were calculated as
(ΔOCR) = (OCR following drug injection − OCR at baseline) and
(ΔECAR) = (ECAR following drug injection − ECAR at baseline).
To normalize each well to cell numbers, after the run, cells were
incubated with Hoechst (1:2000; no. H1399, Sigma-Aldrich) for 15 min
and fixed with 4% PFA for 15 min. Cells were then washed with PBS, and
all the wells were imaged with a Thunder microscope (Leica, DMi8).
Hoechst signal was thresholded, and the area cover was measured using
the software Fiji (ImageJ, 2.0.0).
Electroporation and FACS separation
Proliferating MPC1 WT and MPC1 cKO NSPCs were tranfected with a plasmid
encoding mCherry fused to Cre-recombinase (7 μg/μl, pCAGmCherryCre)
using a 4D-Nucleofector X Unit EA (no. LZ-AAF-1002X, RUWAG) and the P3
primary cell 4D-Nucleofector X solution (no. LZ-V4XP-3024, RUWAG).
NSPCs were cultured in normal proliferative medium for 48 hours before
being subjected to FACS. Cells were trypsinized and resuspended in
EDTA-Dulbecco’s PBS (no. E8008, Merck Millipore) and kept in ice until
the sorting. mCherry-positive cells were sorted on a MoFlo Astrios EQ
cell sorter (Beckman Coulter) and collected in proliferative medium or
directly frozen on dry ice.
Immunocytochemistry
To visualize MPC1 and mitochondria, cells plated on a glass coverslip
were fixed with 4% PFA and 0.2% glutaraldehyde for 15 min. For all the
other stainings, cells were fixed with 4% PFA for 15 to 20 min. Cells
were washed twice for 10 min with PBS and stored at 4°C before
staining. Cells were blocked for 45 min with blocking buffer containing
0.2% Triton X-100 (no. X100, Sigma-Aldrich) and 3% donkey serum (no.
S30, Sigma-Aldrich) in tris-buffered saline [(TBS), 50 mM tris-Cl (pH
7.4), and 150 mM NaCl] and incubated with the indicated primary
antibodies diluted in blocking buffer at 4°C overnight: mouse-pH3
(1:2000; Abcam, ab14955), rabbit-Ki67 (1:500; Abcam, ab15580),
rabbit-MPC1 (1:100; Sigma-Aldrich, HPA045119), rabbit- TOMM20 (1:500;
Abcam ab186735), and mouse-Map2b (1:500; Sigma-Aldrich, M2320). Cells
were washed three times in TBS for 10 min and incubated with secondary
antibodies in blocking buffer for at least 1 hour at room temperature,
protected from light (Alexa Fluor; donkey anti-mouse 488, donkey
anti-rabbit 488, and donkey anti-rabbit 647, 1:250). Cells were washed
twice for 10 min with TBS. Nuclei were stained with
4′,6-diamidino-2-phenylindole (DAPI, 1:5000; no. D9542, Sigma-Aldrich)
for 5 min and washed twice with TBS. Coverslips were mounted with a
self-made polyvinyl alcohol–DABCO–based mounting medium. To detect EdU,
the manufacturer’s protocol was followed (Click-iT Plus EdU Alexa Fluor
647 imaging kit, no. 15224959, Invitrogen).
Image acquisition and analyses
All images used to assess the ablation of MPC1, the quantification of
MAP2AB, and the mitochondrial shape in MPC1 WT and MPC1 cKO were
acquired with a confocal microscope (Zeiss, LSM780) with a 40× or 63×
objective. Images were analyzed using the software Fiji (ImageJ 2.0.0).
Dentritic length was measured using the plugin “Simple Neurite Tracker”
on Fiji. All images to assess the proliferation were acquired in a
blinded manner with an epifluorescent microscope (Nikon 90i or Leica
DMi8) with a 20× or 40× objective. Ten pictures per coverslip were
taken and analyzed using Fiji (ImageJ 2.0.0) in a blinded manner. For
DAPI area cover, 56 tiles with a 10× objective were acquired with an
epifluorescent microscope (Leica DMi8). Area over was measured in
thresholded images. For mitochondrial analysis, serial z-stacks (0.3
μm) were taken using a confocal microscope (Zeiss, LSM780) with a 63×
objective. Mitochondrial length was measured using the freehand tool in
Fiji, as previously described ([199]21).
RNA-seq and qRT-PCR on cultured NSPCs
For RNA-seq, MPC1 WT and MPC1 cKO NSPCs were tranfected with a plasmid
pCAGmCherryCre and sorted according to mCherry (as described above).
Directly after the sort cells were collected and frozen down before RNA
extraction. RNA was extracted using an RNeasy micro kit (Qiagen)
according to the manufacturer’s protocol. RNA-seq data were generated
and analyzed by Alithea Genomics SA (Switzerland).
For in vitro samples, NSPCs were collected and snap-frozen on dry ice.
RNA was extracted using RNeasy mini kit (no. 74134, Qiagen) according
to the manufacturer’s protocol, followed by complementary DNA synthesis
using a SuperScript IV system (no. 15327696, Invitrogen). qRT-PCR was
performed using PowerSYBR Grenn PCR Master Mix (Thermo Fisher
Scientific, no. 10658255). The following primers were used:
Mpc1: 5′-GAC TAT GTC CGG AGC AAG GA-3′ (forward) and 5′-TAG CAA CAG AGG
GCG AAA GT-3′ (reverse); Mpc2: 5′-TGC TGC CAA AGA AAT TG-3′ (forward)
and 5′-AGT GGA CTG AGC TGT GCT GA-3′ (reverse); Got1: 5′- CTT TAA GGA
GTG GAA AGG TAA C-3′ (forward) and 5′- GAG ATA GAT GCT TCT CGT TG-3′
(reverse); Got2: 5′- TAT CAA AAA TCC CAG AGC AG-3′ (forward) and 5′-ATT
CTT TTT CTT CAC CAC GG -3′ (reverse); Asns: 5′- CCA AGT TCA GTA TCC TCT
CC -3′ (forward) and 5′- TAA TTT GCC ACC TTT CTA GC-3′ (reverse);
Slc1a3: 5′- TCA TCT CCA GTC TCG TCA CA-3′ (forward) and 5′- CAC CAC AGC
AAT GAT GGT AGT A-3′ (reverse); Ppat: 5′-CCA GAT AGT ATG TTT GAA GAC
C-3′ (forward) and 5′-CAC TTT GTT GCA TAT CCC AG -3′ (reverse); Pfas:
5′-TTG AAT GCA CTG GAA ATC TG-3′ (forward) and 5′-CGA TCA TCC ACT AGT
ACA ATT C-3′ (reverse); Paics: 5′-GAA TAG CAA CTG GGT CTT TC-3′
(forward) and 5′-ACT TCA GTC TGC CCT ATA AC-3′ (reverse); Adsl: 5′-GAC
AAG ATG GTG ACA GAA AAG-3′ (forward) and 5′-AAT CTG CTG TTT CTC AAA
GG-3′ (reverse); Atic: 5′-GCA TGG TCT ATG ACC TCT AC-3′ (forward) and
5′-AAT ACC ATC CGA CAC TTC TC-3′ (reverse); Cad: 5′-CTT CTT CAG TGT CCA
GTT TC-3′ (forward) and 5′-CTA GAA ATA CAT CGA AGA GCA G-3′ (reverse);
Umps: 5′-AAA AGC AGT ATG AAA GTG GC-3′ (forward) and 5′-CTG TAA GCC TTT
CAC AAC TC-3′ (reverse); and β-actin: 5′-GCC CTG AGG CTC TTT TCC AG-3′
(forward) and 5′-TGC CAC AGG ATT CCA TAC CC-3′ (reverse). β-actin was
used as a housekeeping gene to normalize the obtained cycle threshold
(CT) values, and each sample gene expression was calculated using the
comparative ΔΔCT method.
qRT-PCR on brain tissue
Total RNA from hippocampal tissue was isolated using TRIzol reagent
(Thermo Fisher Scientific) method, and RNA concentration was determined
using NanoDrop One (Thermo Fisher Scientific). Reverse transcription
was performed with 500 to 1000 ng of deoxyribonuclease-treated
(Promega, RQ1 M610A) total RNA using GoScript reverse transcriptase
(Promega A6001). The qRT-PCR was done on a Bio-Rad CFX Connect optics
module. CT values were normalized to β-actin.
Mpc1: 5′-GAC TAT GTC CGG AGC AAG GA-3′ (forward) and 5′-TAG CAA CAG AGG
GCG AAA GT-3′ (reverse); Mpc2: 5′-TGC TGC CAA AGA AAT TG-3′ (forward)
and 5′-AGT GGA CTG AGC TGT GCT GA-3′ (reverse); β-actin:
5′-GGCTGTATTCCCCTCCATCG-3′ (forward) and 5′-CCAGTTGGTAACAATGCCATGT-3′
(reverse).
Protein extraction and Western blotting
Hippocampal brain and cellular extracts were homogenized in
radioimmunoprecipitation assay lysis buffer [50 mM tris-HCl (pH 7.5),
150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1 SDS, 1 mM EDTA,
10% glycerol, and protease inhibitors] and resolved by
SDS–polyacrylamide gel electrophoresis in 8 to 15% polyacrylamide gels.
The concentration of the proteins was determined by Bradford protein
assay using bovine serum albumin as standard and analyzed by Western
blotting with specific antibodies.
Antibodies used were as follows: rabbit-MPC1 (Sigma-Aldrich,
HPA045119), mouse-MPC2 (Merck, MABS1914-25UG), goat-voltage dependent
anion channel protein (VDAC) (Santa Cruz Biotechnology, sc-8829),
mouse-HSP70 (Invitrogen, MA3-028), total OXPHOS antibody cocktail
(Abcam, MS604-300), anti–immunoglobulin G (IgG)–rabbit–horseradish
peroxidase (HRP) (Dako, P0217), anti–IgG-mouse-HRP (Dako, P0447), and
anti–IgG-Goat-HRP (Santa Cruz Biotechnology, sc-2304).
Tissue preparation and immunohistochemistry
All experimental mice were deeply anesthetized with sodium
pentobarbitone (6 mg/100 g body weight, ip) at different time points
after tamoxifen injection (P33, P38, P60, and 10 to 12 months) and
immediately perfused intracardiacally with fresh 4% PFA in 0.1 M PBS
(pH 7.4). Brains were postfixed overnight at 4°C. Coronal sections of
50 and 70 μm were cut using a vibratome (Leica) and stored at 4°C in 1×
PBS supplemented with 0.02% sodium azide. For immunohistochemical
analysis, sections were permeabilized for 1 hour in PBS containing 0.3%
Triton X-100 and 10% donkey or goat serum and then immunolabeled
overnight at 4°C on an orbital shaker, using the following primary
antibodies: mouse-NeuN (1:200; Merck-Millipore, MAB377), chicken-GFP
(1:500; Abcam, ab13970), rabbit-DCX (1:400; Cell Signaling Technology,
4604), guinea pig-DCX (1:100; Merk-Millipore, AB2253), and rabbit-Ki67
(1:400; Abcam, ab16667). For rabbit-MPC1 (1:200; Sigma-Aldrich,
HPA045119) and rat-Ki67 (1:400; Thermo Fisher Scientific, 14-5698-82)
staining, the antibodies were incubated for 72 hours at 4°C. After the
primary antibody incubation, the sections were washed again three times
in PBS for 5 min and incubated for 2 hours at room temperature with
fluorescent secondary antibodies (Alexa Fluor, goat anti-mouse 647;
donkey anti-mouse 647, goat anti-rabbit 488, donkey anti-rabbit 488,
and donkey anti-guinea pig 647; donkey anti-goat 488, donkey
anti-chicken 488, and donkey anti-rat 647, 1:300) diluted in PBS. Last,
nuclei were counterstained with DAPI (1:10,000; Invitrogen) and then
washed before mounting with a FluorSave reagent (Merck-Millipore).
Metabolomics analysis
For untargeted metabolic profiling, adherent cells were washed with a
solution containing 75 mM ammonium carbonate (no. 379999-10G,
Sigma-Aldrich), and the pH was buffered at 7.4 using acetic acid. After
removing the liquid, cells were frozen by holding the place on dry ice.
Extraction of metabolites was performed by adding ice-cold acetonitrile
(no. 34967, Sigma-Aldrich), methanol (no. 34966, Sigma-Aldrich), and
water (no. W6-500, Thermo Fisher Scientific) (4:4:2, v/v) solvent
mixture to cells. The extracts were centrifuged at 13,000 rpm for 2 min
before analysis.
Nontargeted metabolomics analysis was performed by
flow-injection–time-of-flight mass spectrometry (MS) on an Agilent 6550
QTOF system (PMID21830798). The instrument was set to scan in full MS
at 1.4 Hz in negative ionization, 4-GHz high-resolution mode, from 50
to 1000 mass/charge ratio. The solvent was 60:40 isopropanol/water
supplemented with 1 mM NH[4]F at pH 9.0, as well as 10 nM hexakis (1H,
1H, 3H-tetrafluoropropoxy) phosphazine and 80 nM taurochloric acid for
online mass calibration. The injection sequence was randomized. Data
were acquired in profile mode, centroided, and analyzed with Matlab
(The Mathworks, Natick). Missing values were filled by recursion in the
raw data. Upon identification of consensus centroids across all
samples, ions were putatively annotated by accurate mass and isotopic
patterns. Starting from the HMDB v4.0 database, we generated a list of
expected ions, including deprotonated, fluorinated, and all major
adducts found under these conditions. All formulas matching the
measured mass within a mass tolerance of 0.001 Da were enumerated. As
this method does not use chromatographic separation or in-depth MS2
characterization, it is not possible to distinguish between compounds
with identical molecular formula. Principal components analysis and
statistical comparison between the groups were performed using RStudio.
Pathway enrichment analysis was performed using MetaboAnalyst 5.0.
Virus injections in vivo
One-month-old (P30) recombined MPC1cKO-tdTom and control MPC1 WT-tdTom
mice were anesthetized using isoflurane at 5% (w/v), placed in a small
animal stereotaxic frame (David Kopf Instruments), and maintained at
2.5% isoflurane (w/c) for the duration of surgery. Corneal and pinch
reflexes were regularly tested to confirm anesthetic depth. Lacryvisc
(Aicon, Switzerland) was applied to prevent corneal drying and
lidocaine applied topically to the skin overlying the skull. After
exposing the skull under aseptic conditions, a small hole was drilled
in the skull overlying the DG. RV Cre-GFP was injected monolaterally
into the DG from the bregma (anterior posterior, −2 mm; lateral +1.5
mm, 2.3 dorsal ventral mm from skull) at a rate of 100 nl/min^−1
(1.5-μl total volume) using a Hamilton syringe and a CMA400 pump (CMA
System, Kista, Sweden). Mice were euthanized 4 weeks after virus
injections.
Confocal microscopy acquisition and image analysis
All images were collected on a Leica confocal imaging system (TCS SP8)
with a 20× or 63× oil immersion objective, Leica Thunder imaging system
(DMi8) with a 20× objective, and Spinning Disk confocal imaging system
(Nikon Ti2/Yokogawa CSU-W1) with 60× and 100× oil immersion objectives.
For the quantification of recombined tdTom-positive cells with specific
markers, serial sections starting from the beginning of the DG were
used. Quantification of tdTom-positive and Ki67-positive cell labeling
with different markers, DCX and NeuN, was performed using the software
Fiji (ImageJ 2.0.0). To confirm Ki67, DCX, and NeuN colocalization with
tdTom, single optical sections at 63× magnification were used.
For MPC1 quantification in tissue, z-stacks of single Nestin-GFP cells
positive or negative for Ki67 (63× objective) were acquired.
Quantification of MPC1 and GFP signal was performed using IMARIS
software (Bitplane 9.9.1). First, the GFP volume reconstruction was
performed using surface plugin. To confirm the signal of MPC1 in
GFP-expressing cells, MPC1 signal was masked in GFP-positive cells
using mask plugin. Last, MPC1 volume reconstruction of masked signal
was performed using surface plugin. Eighteen to 26 GFP-positive cells
were analyzed per group.
For Sholl analysis, the Cre virus–transfected GFP and tdTom
double-positive newborn neurons were imaged with a 63× [0.75 numerical
aperture (NA)] objective. Z-stacks were taken at 1-μm intervals, and
dendrites were traced using the Neurolucida software (version 10, mbs
Bioscience). Fifteen to 18 neurons from four mice per group were
analyzed. Dendritic spine density and spine morphology were assessed as
previously described ([200]69, [201]70) with little adjustments.
Briefly, dendrites of 20 to 30 Cre virus–transfected GFP and tdTom
double-positive newborn neurons were imaged using a 63× (2.5 NA)
objective. The dendritic length and the number of spines were analyzed
using the software Fiji (ImageJ 2.0.0). Spine density was expressed as
the number of the spines divided by dendritic length. Spine morphology
was classified in two groups on the basis of the maximal diameter of
the spine head, as measured on maximal projections with Fiji (ImageJ
2.0.0). Immature spines (thin spines) were defined as 0.25 to 0.6 μm
and mature spines (mushroom) >0.6 μm. The percentage of each type of
dendritic spine was then expressed by neuron for each mouse (20 to 30
neurons per mouse, four mice per group). The data were expressed as the
ratio between immature spines and mature spines. All images were
analyzed in a blinded manner.
Statistical analyses
Normality of data was tested using the Shapiro-Wilk test. When samples
followed normal distribution, unpaired Student’s t test was conducted
for comparison of two samples, else nonparametric Mann-Whitney test was
used ([202]Fig. 2E and fig.S2D). For metabolomics data, P value was
adjusted for multiple comparisons. For comparisons with more than two
groups, one-way analysis of variance (ANOVA) was performed followed by
Holm-Sidak’s multiple comparison tests. When samples did not follow
normality test, Kruskal-Wallis test was computed. For dendritic
complexity in Sholl analysis, area under the curve was calculated
followed by unpaired Student’s t test. All analyses were performed
using GraphPad Prism 8.0.2 software.
Illustration software
For illustration schemes, Biorender software (Biorender, 2021) and
Adobe Illustrator (version 25.0, Adobe Inc.) were used.
Acknowledgments