Disrupted de novo pyrimidine biosynthesis impairs adult hippocampal neurogenesis and cognition in pyridoxine-dependent epilepsy

Abstract

Despite seizure control by early high-dose pyridoxine (vitamin B6) treatment, at least 75% of pyridoxine-dependent epilepsy (PDE) patients with ALDH7A1 mutation still suffer from intellectual disability. It points to a need for additional therapeutic interventions for PDE beyond pyridoxine treatment, which provokes us to investigate the mechanisms underlying the impairment of brain hemostasis by ALDH7A1 deficiency. In this study, we show that ALDH7A1-deficient mice with seizure control exhibit altered adult hippocampal neurogenesis and impaired cognitive functions. Mechanistically, ALDH7A1 deficiency leads to the accumulation of toxic lysine catabolism intermediates, α-aminoadipic-δ-semialdehyde and its cyclic form, δ-1-piperideine-6-carboxylate, which in turn impair de novo pyrimidine biosynthesis and inhibit NSC proliferation and differentiation. Notably, supplementation of pyrimidines rescues abnormal neurogenesis and cognitive impairment in ALDH7A1-deficient adult mice. Therefore, our findings not only define the important role of ALDH7A1 in the regulation of adult hippocampal neurogenesis but also provide a potential therapeutic intervention to ameliorate the defective mental capacities in PDE patients with seizure control.

ALDH7A1 regulates adult hippocampal neurogenesis and cognition through modulating pyrimidine metabolism.

INTRODUCTION

Pyridoxine-dependent epilepsy (PDE; MIM #266100) is a rare autosomal recessively inherited metabolic disease that is caused by mutations in ALDH7A1, a gene encoding antiquitin, an enzyme central to lysine degradation (1). PDE is characterized by intractable and recurrent neonatal seizures resistant to conventional antiepileptic drugs but responsive to high doses of pyridoxine (vitamin B6) (1, 2). ALDH7A1 deficiency results in the accumulation of toxic lysine catabolism intermediates: α-aminoadipic-δ-semialdehyde (AASA) and its cyclic form, δ-1-piperideine-6-carboxylate (P6C) in blood, urine, and cerebrospinal fluid. Accumulated P6C inactivates pyridoxal 5′-phosphate, the active form of pyridoxine, by forming a Knoevenagel condensation product (1, 3). Despite seizure control with early high-dose pyridoxine treatment, approximately 75% of patients with PDE still suffer from intellectual disability and/or developmental delay (1). Clearly, pyridoxine treatment does not remedy the aberrant accumulation of toxic lysine metabolites (AASA/P6C) in PDE. Therefore, there is a need for additional therapeutic interventions for PDE beyond pyridoxine treatment, which provokes us to study the mechanisms underlying the impairment of brain homeostasis by ALDH7A1 deficiency with seizure control.

Adult hippocampal neurogenesis is a unique process in the mammalian brain, by which new functional neurons are generated from a pool of resident adult neural stem cells (NSCs) (4, 5). Adult hippocampal neurogenesis is important for brain homeostasis and plasticity and is implicated in learning/memory and mood regulation (6, 7). Most NSCs are maintained in a reversible quiescent state, but they exit from quiescence and become reactivated to differentiate into new neurons in response to extrinsic stimuli (8). A reversible switch between activated and quiescent status is essential to maintain the long-lived pool of NSCs and sustain brain homeostasis and functionality throughout the life span (9). Besides regulating by a plethora of morphogenic signaling and transcriptional factors (5, 9, 10), recent evidence has demonstrated that amino acids and their derived metabolites are involved in cellular metabolism during adult neurogenesis, and dysregulation of amino acid metabolism impairs adult neurogenesis and cognition (11–13). Therefore, deciphering the mechanisms underlying the regulation of adult hippocampal neurogenesis by amino acids and their derived metabolites may pave the way to develop potential therapeutic strategies to ameliorate the mental health in patients with inborn errors of amino acid metabolism or aminoacidopathies, including PDE.

RESULTS

Brain-specific deletion of ALDH7A1 leads to pyridoxine-dependent epilepsy

To investigate the role of ALDH7A1 in the brain, we generated Aldh7a1 conditional knockout (Aldh7a1-cKO) mice by crossing Aldh7a1^f/f mice with Nestin-Cre transgenic mice (fig. S1A) and maintained them on the facility’s standard chow (referred to as regular diet), which contains 1.3% lysine and 13 parts per million (ppm) of pyridoxine. The ALDH7A1 expression was completely ablated from the brain of Aldh7a1-cKO mice (fig. S1B). However, Aldh7a1-cKO mice displayed similar growth rates as the wild-type (WT) littermates (fig. S1, C and D). Adult Aldh7a1-cKO mice were indistinguishable from their WT littermates in size, appearance, and home cage behavior. In addition, the brain weight and the ratio of brain weight to body weight were comparable between Aldh7a1-cKO and WT mice (fig. S1, E to G).

To test whether spontaneous seizures occur in Aldh7a1-cKO mice, we performed multichannel electroencephalogram (EEG) recording for 72 hours at 2 months of age. There was no evidence of convulsive seizure in Aldh7a1-cKO mice when feeding with a regular diet (fig. S1, H and I, and table S1). We hypothesized that the pyridoxine levels in the diet might affect the seizure susceptibility of Aldh7a1-cKO mice. According to the National Research Council, 1995, pyridoxine at 1 ppm in the diet is adequate for the growth and maintenance of adult mice. We then subjected the mice to the special diet (SD) that contains 1.3% lysine and 1.6 ppm of pyridoxine and performed multichannel EEG recording for 72 hours at 2 months of age. WT mice did not exhibit convulsive seizure on SD (fig. S1, H and I, and table S1). However, all Aldh7a1-cKO mice displayed spontaneous electrographic seizures when feeding with SD (fig. S1, H and I, and table S1). Furthermore, Aldh7a1-cKO mice began to die on day 2, but only ~60% of Aldh7a1-cKO mice could survive when following up the SD trial for 30 days (fig. S1J). Therefore, specific deletion of ALDH7A1 from the brain is sufficient to induce spontaneous seizures, which is highly dependent on dietary pyridoxine levels.

Catabolism of lysine to aminoadipate through the saccharopine pathway is impaired in the brain of Aldh7a1-cKO mice

Lysine is catabolized in mammals through the saccharopine and pipecolate pathways, both of which lead to the formation of AASA and then are oxidized to aminoadipate (AAA) by ALDH7A1 (Fig. 1A) (14). However, a distinctive feature of these two pathways is the differential deamination of nitrogen from lysine (15, 16). In the saccharopine pathway, the lysine’s nitrogen epsilon (ε-N) is deaminated and its nitrogen alpha (α-N) remains in the downstream metabolites, AASA/P6C and AAA (Fig. 1A). Since pipecolate can be produced from P6C, pipecolate could be derived from the ε-deaminated lysine (Fig. 1A). In the pipecolate pathway, the lysine α-N is removed from lysine, and the ε-N is retained in the downstream pipecolate, which is further oxidized and persists the lysine-derived ε-N in AASA/P6C and AAA (Fig. 1A).

Fig. 1. Aldh7a1-cKO mice display dysregulation of lysine metabolites and impaired cognitive functions.

(A) Schematic illustration of the saccharopine or pipecolic acid pathways of nitrogen deamination from lysine by tracing α-¹⁵N-lysine or ε-¹⁵N-lysine. (B) Mass spectrometry analysis of ¹⁵N-labeled saccharopine, pipecolic acid, AASA, P6C, and AAA from the catabolism of α-¹⁵N-lysine or ε-¹⁵N-lysine in the brain of WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 3 mice; **P < 0.01 and ***P < 0.001; Student’s t test). (C) Schematic diagram of Y-maze test. (D) Quantification of spontaneous alternations between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 12 mice; **P < 0.01; Student’s t test). (E) Schematic diagram of novel object location test. (F) Quantification of exploring time on the novel location of the same object between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 12 mice; ***P < 0.001; Student’s t test). (G) Schematic diagram of novel object recognition test. (H) Quantification of exploring time on the novel object between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 12 mice; **P < 0.01; Student’s t test). (I) Schematic diagram of fear conditioning test. (J) Quantification of freezing behaviors to contextual stimuli between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 12 mice; **P < 0.01; Student’s t test). (K) Quantification of freezing behaviors to cue stimuli between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 12 mice; ***P < 0.001; Student’s t test).

To investigate the lysine catabolism in the brain, we performed liquid chromatography–mass spectrometry in mice injected with lysine labeled at either α-¹⁵N or ε-¹⁵N. Since saccharopine contains both α- and ε-N derived from lysine, ¹⁵N-saccharopine was found in the brain of WT mice injected with both α-¹⁵N-lysine and ε-¹⁵N-lysine (Fig. 1B). ¹⁵N-P6C and ¹⁵N-AASA were barely detected in the brain of WT mice from α-¹⁵N-lysine and ε-¹⁵N-lysine (Fig. 1B), suggesting a fast turnover rate of these two toxic lysine catabolism intermediates in the brain. However, the levels of α-¹⁵N-pipecolate were higher than those of ε-¹⁵N-pipecolate in the brain of WT mice (Fig. 1B), indicating production of pipecolate mostly from saccharopine pathways in the brain. Consistent with a previous study (16), we found that ¹⁵N-AAA was primarily detected in the brain of WT mice from α-¹⁵N-lysine, but not from ε-¹⁵N-lysine (Fig. 1B), suggesting the saccharopine pathway primarily for AAA biosynthesis in the brain.

Next, we assessed the consequences of ALDH7A1 deficiency on lysine metabolism in the brain. Incorporation of α-¹⁵N, but not ε-¹⁵N, from lysine into saccharopine, AASA/P6C, and pipecolic acid was stabstantially increased in the brain of Aldh7a1-cKO mice compared to WT controls (Fig. 1B), suggesting the blockage of lysine catabolism via saccharopine pathway upon ALDH7A1 deletion. Unexpectedly, we observed an increase in α-¹⁵N-AAA, a downstream metabolite of lysine catabolism by ALDH7A1, in the brains of Aldh7a1-cKO mice (Fig. 1B). Since the loss of ALDH7A1 activity should inhibit the production of AAA, we anticipated that AAA levels would be decreased by ALDH7A1 deficiency. This observed increase in AAA suggests that there is an alternate route of AAA production from lysine in the brain independent of saccharopine and pipecolate pathways (17, 18).

Together, the above data suggest that ALDH7A1 deficiency leads to the blockage of lysine catabolism through the saccharopine pathway, thereby accumulating toxic lysine metabolites (AASA/P6C) in the brain.

Aldh7a1-cKO mice show impaired cognitive performance

To assess whether the cognitive functions were altered in Aldh7a1-cKO mice with seizure control on a regular diet, we subjected the mice to a battery of behavior tests. Compared with the WT littermates, Aldh7a1-cKO mice exhibited reduced alternation rates during the Y-maze test (Fig. 1, C and D), indicating a deficit in short-term memory. In addition, Aldh7a1-cKO mice displayed less preference for the novel location of the same object during the novel object location (NOL) test (Fig. 1, E and F), suggesting a deficit in spatial memory and discrimination. Similarly, in the novel objection recognition (NOR) test, Aldh7a1-cKO mice showed a reduced exploratory preference for novel objects (Fig. 1, G and H). Moreover, Aldh7a1-cKO mice showed significantly impaired associative learning, as shown by reduced freezing behavior to contextual and cue stimuli in the fear conditional test (Fig. 1, I to K). However, there was no significant difference between WT and Aldh7a1-cKO mice during the open field and elevated plus maze tests (fig. S2), suggesting that the deficits in learning and memory of Aldh7a1-cKO mice result from cognitive impairment rather than locomotor activity or anxiety.

Aldh7a1-cKO mice display impaired hippocampal neurogenesis

Adult hippocampal neurogenesis is essential to preserve learning and memory-related cognitive functions (6, 7). To explore the role of ALDH7A1 in adult hippocampal neurogenesis, we examined the expression pattern of Aldh7A in the dentate gyrus (DG) of the adult hippocampus. ALDH7A1 was expressed in Sox2⁺GFAP⁺ radial-like NSCs, Tbr2⁺ intermediate progenitor cells (IPCs), and Sox2⁺GFAP⁺ star-like astrocyte, but barely in DCX⁺ neuroblasts and NeuN⁺ mature neurons (MNs) (Fig. 2A) in the adult DG.

Fig. 2. Aldh7a1-cKO mice exhibit impaired hippocampal neurogenesis.

(A) Representative images for ALDH7A1 expression costained with GFAP, Sox2, Tbr2, DCX, or NeuN in the adult DG. Scale bars, 20 μm. (B) Experimental scheme for assessing cell proliferation by a 2-hour EdU pulse-chase in P60 mice. (C) Representative images of the brain sections from WT and Aldh7a1-cKO mice stained with EdU, Sox2, and GFAP. Scale bars, 20 μm. (D and E) Quantification of GFAP⁺Sox2⁺EdU⁺ NSCs and GFAP⁻Sox2⁺EdU⁺ progenitors in the adult DG of WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 5 mice; *P < 0.05 and ***P < 0.001; Student’s t test). (F) Representative images of Dcx⁺ cells in the brain of WT and Aldh7a1-cKO mice. Scale bars, 50 μm. (G) Quantification of Dcx⁺ neuroblasts in the adult DG of WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 5 mice; ***P < 0.001; Student’s t test). (H) Experimental scheme for assessing adult-born neurons by a 30-day EdU pulse-chase in P60 mice. (I) Representative images of brain sections from WT and Aldh7a1-cKO mice stained with NeuN and EdU. Scale bars, 20 μm. (J) Quantification of NeuN⁺EdU⁺ adult-born neurons in the adult DG of WT and Aldh7a1-cKO mice at P90. Values represent means ± SEM (n = 5 mice; **P < 0.01; Student’s t test).

Next, we determined whether adult hippocampal neurogenesis was affected in Aldh7a1-cKO mice with seizure control on a regular diet. The mice were injected 5-ethynyl-2′-deoxyuridine (EdU) at postnatal day 60 (P60) and the brains were collected for analysis 2 hours later (Fig. 2B). Quantification showed significant decreases in the number of EdU⁺GFAP⁺Sox2⁺ NSCs and EdU⁺GFAP⁻Sox2⁺ progenitors in Aldh7a1-cKO mice (Fig. 2, C to E). Moreover, by using Dcx to label the neuroblasts, we found that the number of Dcx⁺ cells was markedly decreased in Aldh7a1-cKO mice (Fig. 2, F and G). To examine the production of adult-born MNs, we injected P60 mice with EdU four times at 12-hour intervals and collected brain tissues 30 days later (Fig. 2H). There were significant decreases in the number of EdU⁺NeuN⁺ adult-born MNs in Aldh7a1-cKO mice (Fig. 2, I and J). Thus, these results suggest that adult hippocampal neurogenesis is impaired in ALDH7A1-deficient mice even with seizure control.

Specific deletion of ALDH7A1 from adult NSCs inhibits NSC activation and subsequent neurogenesis

Given that ALDH7A1 is expressed by both NSCs and astrocytes in the adult DG, we then examine the specific contribution of ALDH7A1 from NSCs on adult neurogenesis. We generated the Aldh7a1^f/f::Nestin-CreER^T2::Ai14 inducible knockout mouse model (Aldh7a1-iKO) by crossing Aldh7a1^f/f conditional allele with the Nestin-CreER^T2 driver and the Rosa26-tdTomato reporter (Ai14) mouse lines (fig. S3A). The expression of ALDH7A1 was almost undetectable in tdTomato-positive (tdT⁺) NSCs of Aldh7a1-iKO mice after tamoxifen-induced recombination (fig. S3B). By performing multi-channel EEG recording for 72 hours at 30 days after tamoxifen injection, we found that there was no evidence of convulsive seizure in adult Aldh7a1-iKO mice when feeding either with regular diet or with SD (fig. S3, C and D, and table S1). These results suggest that specific deletion of ALDH7A1 from adult NSCs does not contribute to spontaneous seizures.

To assess the specific effect of ALDH7A1 deficiency on NSCs in the adult hippocampus, we injected the mice with tamoxifen for analysis 2 days later (Fig. 3A). Quantification showed that the numbers of tdT⁺ cells and tdT⁺Sox2⁺GFAP⁺ NSCs were comparable between WT and Aldh7a1-iKO mice (Fig. 3, B to D). However, by using a cell activation marker (Mcm2) to distinguish between quiescent and activated NSCs (aNSCs), we found that there were fewer tdT⁺GFAP⁺Mcm2⁺ aNSCs in Aldh7a1-iKO mice than in WT controls (Fig. 3, B and E), indicating an excessive persistence of NSCs in quiescent status upon loss of ALDH7A1.

Fig. 3. Specific deletion of ALDH7A1 from adult NSCs leads to inhibited NSC activation.

(A) Experimental scheme for assessing NSC activation after a 2-day post-tamoxifen induction in P60 mice. (B) Representative images of the brain sections from WT and Aldh7a1-iKO mice stained with Mcm2 and GFAP. Scale bars, 20 μm. (C) Quantification of tdT⁺ cells in the adult DG of WT and Aldh7a1-iKO mice after a 2-day post-tamoxifen induction. Values represent means ± SEM (n = 5 mice). (D) Quantification of tdT⁺GFAP⁺ NSCs in the adult DG of WT and Aldh7a1-iKO mice after a 2-day post-tamoxifen induction. Values represent means ± SEM. (E) Percentage of activated NSCs (aNSCs; tdT⁺GFAP⁺Mcm2⁺) or quiescent NSCs (qNSCs; tdT⁺GFAP⁺Mcm2⁻) among total NSCs (tdT⁺GFAP⁺) in the adult DG of WT and Aldh7a1-iKO mice. Values represent means ± SEM (n = 5 mice). (F) Experimental scheme for assessing cell cycle exit of NSCs after a 2-day post-tamoxifen induction and followed by a 24-hour EdU pulse labeling in P60 mice. (G) Representative images of the brain sections from WT and Aldh7a1-iKO mice stained with EdU and Ki67. Scale bars, 20 μm. (H) Percentage of aNSCs that exited cell cycle (tdT⁺EdU⁺Ki67⁻) or re-entered cell cycle (tdT⁺EdU⁺Ki67⁺) among total NSCs (tdT⁺EdU⁺) in the adult DG of WT and Aldh7a1-iKO mice. Values represent means ± SEM (n = 5 mice).

To assess whether the loss of ALDH7A1 promotes NSCs return to quiescence, we labeled a cohort of proliferating NSCs with a pulse of EdU at 2 days after tamoxifen injection and identified the proportion of NSCs that had exited the cell cycle 24 hours later by evaluating the labeling for EdU and Ki67 (Fig. 3F). Quantification showed that there were more tdT⁺EdU⁺Ki67⁻ NSCs, which had exited the cell cycle, in Aldh7a1-iKO mice than in WT controls (Fig. 3, G and H). Hence, these data indicate that ALDH7A1 deficiency promotes NSCs exit from the cell cycle and drives aNSCs into quiescence.

Once activated, the majority of NSCs undergo limited proliferative activity, then differentiate, and eventually become exhausted (19, 20). To examine the long-term consequences of specific deletion of ALDH7A1 from NSCs on adult neurogenesis, we injected P60 mice with multiple doses of tamoxifen for analysis 30 days later (Fig. 4A). Quantification showed significantly fewer tdT⁺ cells in Aldh7a1-iKO mice than in WT controls (Fig. 4, B and C). By using cell-lineage markers to map the fate of tdT⁺ cells, we found fewer tdT⁺Tbr2⁺ IPCs, tdT⁺DCX⁺ neuroblasts, and tdT⁺NeuN⁺ MNs, but more tdT⁺GFAP⁺ NSCs, in Aldh7a1-iKO mice than in WT mice (Fig. 4, D and E). Therefore, these data corroborate our claim that ALDH7A1 deficiency drives aNSCs into quiescence, thereby inhibiting the production of newborn neurons in the adult hippocampus.

Fig. 4. Specific deletion of ALDH7A1 from adult NSCs leads to impaired hippocampal neurogenesis and hippocampus-dependent cognition.

(A) Experimental scheme for assessing NSCs and their progeny after a 30-day post-tamoxifen induction in P60 mice. (B) Representative images of tdT⁺ cells in the adult DG of WT and Aldh7a1-iKO mice. Scale bars, 20 μm. (C) Quantification of tdT⁺ cells in the adult DG of WT and Aldh7a1-iKO mice after a 30-day post-tamoxifen induction. Values represent means ± SEM (n = 5 mice; **P < 0.01; Student’s t test). (D) Representative images used for fate mapping of tdT⁺ cells in the DG by costaining with GFAP (NSCs), Tbr2 (IPCs), DCX (INs), and NeuN (MNs). Scale bars, 20 μm. (E) Quantification of the composition of tdT⁺ cells in the adult DG of WT and Aldh7a1-iKO mice. Values represent means ± SEM (n = 5 mice; *P < 0.05, **P < 0.01, and ***P < 0.001; Student’s t test). (F) Experimental scheme for behavioral tests after a 30-day post-tamoxifen induction in P60 mice. (G) Quantification of exploring time on the novel location of the same object between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 10 mice; *P < 0.05; Student’s t test). (H) Quantification of exploring time on the novel object between WT and Aldh7a1-iKO mice. Values represent means ± SEM (n = 10 mice; *P < 0.05; Student’s t test). (I) Quantification of freezing behaviors to contextual stimuli between WT and Aldh7a1-cKO mice. Values represent means ± SEM (n = 10 mice; **P < 0.01; Student’s t test).

Specific deletion of ALDH7A1 from adult NSCs leads to impaired hippocampus-dependent learning

To assess whether hippocampus-dependent learning capacity is affected after specific deletion of ALDH7A1 from adult NSCs, we injected the mice with multiple doses of tamoxifen at P60 and subjected them to hippocampus-dependent learning tests (21, 22), including NOL, NOR, and contextual fear conditioning test, after 30 days of tamoxifen injection (Fig. 4F). As expected, Aldh7a1-iKO mice displayed less preference to the objects in both NOL and NOR tests compared to WT controls (Fig. 4, G and H). Moreover, Aldh7a1-iKO mice exhibited reduced freezing behavior to contextual stimuli during fear conditioning test (Fig. 4I). These data suggest that the defective adult hippocampal neurogenesis might directly contribute to the cognitive impairments in ALDH7A1-deficient mice even with seizure control.

The accumulation of AASA/P6C by ALDH7A1 deficiency inhibits NSC proliferation and differentiation

To investigate the mechanism underlying the regulation of neurogenesis by ALDH7A1-mediated lysine catabolism, we isolated adult neural stem/progenitor cells from Aldh7a1^f/f mice (Aldh7a1^f/f-NPCs) and maintained them in regular culture medium (containing 30 nM pyridoxine, referred to as regular medium). To induce Aldh7a1 gene deletion, we transduced the cells with lentivirus coexpressing green fluorescence protein (GFP) with Cre recombinase (lenti-CreGFP) or truncated Cre lacking recombinase activity (lenti-dCreGFP) (fig. S4A). Compared to the cells infected with lenti-dCreGFP (referred to as WT-NPCs), infection with lenti-CreGFP led to ablation of ALDH7A1 expression in Aldh7a1^f/f-NPCs (referred to as KO-NPCs) (fig. S4B). KO-NPCs exhibited reduced cell proliferation and neuronal differentiation compared to WT-NPCs either in a regular medium or in a pyridoxine-free medium (fig. S4, C and D). Moreover, the defective cell proliferation and neuronal differentiation of KO-NPCs could not be restored by administration of pyridoxine in the therapeutic concentration (23) (3000 nM) for seizure control in PDE or even above the therapeutic concentration (>3000 nM) (fig. S4, C and D). In addition, both WT- and KO-NPCs did not exhibit an elevation of caspase-3 activation (fig. S4E), excluding the involvement of apoptosis in this process. Thus, pyridoxine administration is not able to rescue defective proliferation and differentiation of ALDH7A1-deficient NSCs, which is in agreement with in vivo observations.

To examine whether the accumulation of downstream metabolites (saccharopine, AASA/P6C, pipecolic acid, and AAA) via lysine catabolism is accountable for the impairment of NSC proliferation and differentiation, we treated WT-NPCs with each of the metabolites and performed NPC proliferation and differentiation assays in vitro (Fig. 5A and fig. S4F). The provision of AASA/P6C (Fig. 4, B to E), but not saccharopine (fig. S4, G and H), pipecolic acid (fig. S4, J and K), or AAA (fig. S4, M and N), inhibited NPC proliferation and neuronal differentiation. Nonetheless, no elevation of caspase-3 activation was observed after the provision of these metabolites (Fig. 5, F and G, and fig. S4, I, L, and O), excluding the involvement of apoptosis during this treatment. Therefore, these data suggest that the accumulation of AASA/P6C by ALDH7A1 deficiency results in impaired NSC proliferation and differentiation.

Fig. 5. Accumulated AASA/P6C leads to impaired NSC proliferation and differentiation.

(A) Schematic illustration for assessing the effect of lysine downstream metabolite (AASA/P6C) in NPCs in vitro. (B) Representative images of EdU⁺ NPCs treated with or without AASA/P6C. Scale bars, 20 μm. (C) Percentage of EdU⁺ cells among total DAPI⁺ cells upon AASA/P6C treatment. Values represent means ± SEM (n = 3 independent experiments with three technical replicates; **P < 0.01; Student’s t test). (D) Representative images of Tuj1⁺ differentiated neurons from NPCs treated with or without AASA/P6C. Scale bars, 20 μm. (E) Percentage of Tuj1⁺ cells among total DAPI⁺ cells upon AASA/P6C treatment. Values represent means ± SEM (n = 3 independent experiments with three technical replicates, **P < 0.01; Student’s t test). (F) Representative images of AC3⁺ NPCs treated with or without AASA/P6C. Scale bars, 20 μm. (G) Percentage of AC3⁺ NPCs among total DAPI⁺ NPCs upon AASA/P6C treatment. Values represent means ± SEM (n = 3 independent experiments with three technical replicates).

Transcriptome analyses reveal impaired pyrimidine metabolism in ALDH7A1-deficient NSCs

To explore the molecular mechanism underlying dysregulation of NSC proliferation and differentiation by ALDH7A1 deficiency, we performed RNA sequencing (RNA-seq) on WT- and KO-NPCs that were cultured in pyridoxine (3000 nM) or pyridoxine-free medium. Using a sample-to-sample distance heatmap to visualize overall gene expression between samples, we found that the WT sample was clustered with the WT + pyridoxine sample, whereas the KO sample was grouped with the KO + pyridoxine sample (fig. S5A). Moreover, principal components analysis of the whole transcriptome demonstrated that WT and KO samples were distinctly separated, which was independent of pyridoxine treatment (fig. S5B). Furthermore, the statistical analyses of differentially expressed genes (DEGs) showed significant DEGs between WT and KO samples, which were barely affected by pyridoxine levels (fig. S5C). Thus, these whole transcriptome analyses support our conclusion that supplementation of pyridoxine has minimal beneficial effect on ALDH7A1-deficient NSCs.

To compare the DEGs between samples, we found a total of 868 DEGs between WT and KO samples, with 413 up-regulated and 455 down-regulated (fig. S5D and table S2), and a total 903 DEGs between WT + pyridoxine and KO + pyridoxine samples, with 451 up-regulated and 452 down-regulated (Fig. 6A and table S3). Moreover, the KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analyses demonstrated the putative or predicted pathways of the DEGs between WT and KO samples were highly similar to those between WT + pyridoxine and KO + pyridoxine samples (Fig. 6B and fig. S5E). For example, the up-regulated DEGs were enriched in focal adhesion, ECM-receptor interaction, and calcium signaling pathway, while the down-regulated DEGs were enriched in cell cycle, DNA replication, and pyrimidine metabolism.

Fig. 6. Transcriptome analysis reveals impaired pyrimidine biosynthesis in ALDH7A1-deficient NSCs.

(A) The heatmap of expression levels of DEGs in WT- and KO-NPCs with pyridoxine treatment and their replicates. (B) KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis of DEGs between WT- and KO-NPCs with pyridoxine treatment. (C) The heatmap of expression levels of genes involved in pyrimidine metabolism between WT- and KO-NPCs after pyridoxine treatment. (D) qRT-PCR analysis of expression levels of genes involved in pyrimidine metabolism between WT- and KO-NPCs after pyridoxine treatment. Values represent means ± SEM (n = 3 independent experiments with three technical replicates). (E) Schematic illustration of de novo pyrimidine biosynthesis and salvage, the key enzymes, and the down-regulation of enzymes in ALDH7A1-deficient NSCs.

Pyrimidine metabolism is activated in proliferating cells in response to an increased demand for nucleotides needed for DNA synthesis (24). We found a similar subset of down-regulated DEGs related to pyrimidine metabolism in both KO and KO + pyridoxine samples (Fig. 6C and fig. S5F). Moreover, these down-regulated DEGs were further validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR) assay (Fig 6D and fig. S5G). Notably, the provision of AASA/P6C to WT-NPCs led to reduced expression levels of the same set of genes involved in de novo pyrimidine biosynthesis (fig. S5, H and I). Therefore, ALDH7A1 deficiency leads to AASA/P6C accumulation, which in turn impairs de novo pyrimidine biosynthesis in adult NSCs (Fig. 6E).

ALDH7A1 deficiency leads to impaired de novo pyrimidine biosynthesis

Although there are salvage pathways and cells can take up nucleotides, most proliferating cells synthesize nucleotides and nucleic acids de novo, mainly from glucose and glutamine (24). To assess whether de novo pyrimidine biosynthesis is impaired in ALDH7A1-deficient NPCs, we traced stable isotope-labeled ¹³C-glucose and ¹³C-glutamine in WT- and KO-NPCs (Fig. 7A). The labeling of pyrimidine from ¹³C-glucose or ¹³C-glutamine was significantly reduced in KO-NPCs compared to WT controls (Fig. 7, B and C). However, the incorporation of labeled carbon from ¹³C-glucose into the intermediates of glycolysis, pentose phosphate pathway, or tricarboxylic acid cycle was not affected in KO-NPCs (fig. S6, A to D). In addition, the ¹³C incorporation from glutamine into the intermediates via oxidative glutaminolysis was not affected in KO-NPC (fig. S6, E and F). Together, these findings suggest that ALDH7A1 deficiency impairs de novo pyrimidine biosynthesis but does not alter the metabolic flux of glucose or glutamine during energy metabolism.

Fig. 7. ALDH7A1 deficiency leads to impaired de novo pyrimidine biosynthesis from glucose and glutamine.

(A) Schematic illustration of pyrimidine biosynthesis from glucose and glutamine. (B) Quantification of incorporation of ¹³C-glucose in pyrimidines [uridine 5′-monophosphate (UMP), cytidine 5′-monophosphate (CMP), and 2'-deoxyruidine 5'-Monophophate (dTMP)] in WT- and KO-NPCs. Values represent means ± SEM (n = 3 independent experiments with three technical replicates; *P < 0.05 and ***P < 0.001; Two-way ANOVA). (C) Quantification of incorporation of ¹³C-glutamine in pyrimidines (UMP, CMP, and dTMP) in WT- and KO-NPCs. Values represent means ± SEM (n = 3 independent experiments with three technical replicates; **P < 0.01 and ***P < 0.001; Two-way ANOVA).

In line with the critical role of de novo pyrimidine biosynthesis on cell proliferation, we inhibited dihydroorotate dehydrogenase, a rate-limiting enzyme of the de novo synthesis pathway of pyrimidine nucleotides, by leflunomide (25) (fig. S7A). NPCs treated with leflunomide exhibited reduced cell proliferation and neuronal differentiation (fig. S7, B and C). Nevertheless, no elevation of caspase-3 activation was observed after leflunomide treatment (fig. S7D). Although nucleotide depletion induces nuclear DNA damage (26), the DNA damage response pathway is not activated in ALDH7A1-deficient NPCs or leflunomide-treated NPCs, as assessed by phosphorylation of checkpoint kinase 1 (fig. S7, E and F), which is a marker for DNA damage (27). Thus, these observations suggest that disrupted de novo pyrimidine biosynthesis in ALDH7A1-deficient NSCs results in inhibiting cell proliferation and neuronal differentiation but does not induce nuclear DNA damage.

Supplementation of pyrimidines rescues defective adult hippocampal neurogenesis and cognitive impairment in ALDH7A1-deficient mice

Since de novo pyrimidine biosynthesis is diminished upon ALDH7A1 deletion in NSCs, we then investigated whether the exogenous supply of pyrimidines could restore defective cell proliferation and neuronal differentiation caused by ALDH7A1 deficiency. Notably, supplementation of a combination of thymidine, cytidine, and uridine to KO-NPCs was sufficient to promote cell proliferation and neuronal differentiation (fig. S8, A and B), indicating that ALDH7A1-deficient NPCs could be rescued by the exogenous provision of pyrimidines.

The above in vitro data encouraged us to assess whether there is a rescue effect of pyrimidines supplementation on defective hippocampal neurogenesis in Aldh7a1-iKO mice. We infused a combination of thymidine, cytidine, and uridine into adult brain after tamoxifen injection and analyzed 4 weeks later (Fig. 8A). There were significant increases in the number of tdT⁺ cells in Aldh7a1-iKO mice after pyrimidines administration (Fig. 8B). By using cell-lineage markers to map the fate of tdT⁺ cells, we found that supplementation of pyrimidines increased the numbers of tdT⁺Tbr2⁺ IPCs, tdT⁺DCX⁺ neuroblasts, and tdT⁺NeuN⁺ MNs, but reduced the number of tdT⁺GFAP⁺ NSCs, in Aldh7a1-iKO mice (Fig. 8C). Furthermore, by using Mcm2 to distinguish between the quiescent and the activated states of NSCs, we found that the proportion of tdT⁺GFAP⁺Mcm2⁻ quiescent NSCs was decreased in Aldh7a1-iKO mice after treatment with pyrimidines (Fig. 8D).

Fig. 8. Supplementation of pyrimidines ameliorates defective adult hippocampal neurogenesis of Aldh7a1-iKO mice.

(A) Experimental scheme for assessing NSCs and their progeny in the mice by TAM injection and followed by supplementation of pyrimidines. (B) Quantification of tdT⁺ cells in the DG of WT and Aldh7a1-iKO mice treated with vehicle or pyrimidines. Values represent means ± SEM (n = 3 mice; *P < 0.05; Student’s t test). (C) Quantification of NSCs, IPCs, INs, and MNs in the adult DG of WT and Aldh7a1-iKO mice treated with vehicle or pyrimidines. Values represent means ± SEM (n = 3 mice; *P < 0.05 and **P < 0.01; Student’s t test). (D) Percentage of aNSCs (tdT⁺GFAP⁺Mcm2⁺) and qNSCs (tdT⁺GFAP⁺Mcm2⁻) among total NSCs (tdT⁺GFAP⁺) in the adult DG of WT and Aldh7a1-iKO mice treated with vehicle or pyrimidines. Values represent means ± SEM (n = 3 mice). (E) Experimental scheme for assessing cell cycle exit of NSCs in the mice treated with vehicle or pyrimidines. (F) Percentage of NSCs exited cell cycle among total EdU-labeled NSCs in the adult DG of WT and Aldh7a1-iKO mice treated with vehicle or pyrimidines. Values represent means ± SEM (n = 3 mice).

To investigate whether an infusion of pyrimidines could promote Aldh7a1-deficient NSCs to enter the cell cycle, we infused the mice with pyrimidines after tamoxifen injection, marked a cohort of proliferating cells with a pulse of EdU 2 days after pyrimidines treatment, and identified the fraction of NSCs that had exited the cell cycle 24 hours after EdU injection by evaluating the labeling for EdU and Ki67 (Fig. 8E). After pyrimidines treatment, the proportion of EdU⁺Ki67⁻ NSCs that had exited the cell cycle was significantly reduced in Aldh7a1-iKO mice compared to those treated with vehicle (Fig. 8F). These data suggest that supplementation of pyrimidines is able to prevent ALDH7A1-deficient NSCs from returning to quiescence, thereby ensuring proper adult hippocampal neurogenesis.

To assess whether there is a rescue effect of pyrimidines administration on hippocampus-dependent learning in Aldh7a1-iKO mice, we infused the mice with pyrimidines after tamoxifen injection and analyzed 4 weeks later (fig. S9A). Obviously, infusion of pyrimidines significantly ameliorated the defective cognitive performances of Aldh7a1-iKO mice, as shown by improving the freezing behaviors in the fear contextual conditioning test (fig. S9B), as well as promoting the exploring preference for the novel location of the same object in the NOL assessment (fig. S9C) or the novel object in the NOR test (fig. S9D).

Last, we investigated whether pyrimidine supplementation can ameliorate adult neurogenesis and cognitive deficits in Aldh7a1-cKO mice (fig. S10A). Quantification showed that pyrimidine supplementation rescued defective neurogenesis in the adult hippocampus of Aldh7a1-cKO mice, as shown by increasing the number of Dcx⁺ cells (fig. S10B). Furthermore, pyrimidine supplementation was able to ameliorate the cognitive deficits of Aldh7a1-cKO mice, as shown by improving performances in the fear contextual conditioning (fig. S9C), NOL (fig. S10D), and NOR tests (fig. S10E).

DISCUSSION

Understanding how amino acid metabolism regulates NSC proliferation and differentiation not only is an important basis in the field of neurogenesis but also sheds light on brain plasticity and regenerative medicine (11). In this study, we demonstrated that the deficiency of ALDH7A1, a key enzyme in lysine oxidation, led to the accumulation of toxic lysine catabolism intermediates, AASA and its cyclic form, δ-1-piperideine-6-carboxylate, which in turn impaired de novo pyrimidine biosynthesis and inhibited NSC proliferation and differentiation (fig. S11).

Although the discovery of ALDH7A1 mutation as a genetic cause of PDE in 2006 (3), the pathophysiological mechanisms underlying the neurodevelopmental impairments have not yet been completely understood (1). Because of the lack of an appropriate animal model for PDE, it significantly hampers the progress in this field. In 2020, Al-Shekaili et al. (17) developed a mouse model of ALDH7A1 deficiency in C57BL/6 J and 129 genetics background, which recapitulates the biochemical abnormalities and a dietary-induced, pyridoxine-responsive seizure susceptibility of patients with PDE. However, cognitive or neuropathological abnormalities have not been observed in their mouse model (17). In our study, we generated a PDE mouse model in C57BL/6 N genetic background, in which ALDH7A1 was specifically deleted from the central nervous system. Our mouse model not only exhibited the blockade of the lysine catabolism pathway by ALDH7A1 deficiency and a pyridoxine-dependent clinical seizure phenotype but also displayed impaired cognitive functions even with seizure control. Despite the fact that we did not directly compare the accumulated levels of toxic lysine catabolism intermediates (AASA/P6C) between our mouse model and Al-Shekaili et al.’s reported ones (17), there is no obvious correlation between phenotype severity and AASA/P6C levels in patients with PDE (28). On the other hand, small genetic differences between C57BL/6 inbred substrains have a great influence on neural-behavioral phenotypes (29), which may explain the discrepancy between our mouse model and Al-Shekaili et al.’s reported ones on cognitive performances. Note that there is the existence of heterogeneous clinical phenotypes in PDE: patients with complete seizure control upon pyridoxine administration and normal developmental outcomes or seizure control but developmental delay; or those with persistent seizures despite pyridoxine treatment and developmental delay (30). Moreover, patients with PDE normally display neonatal-onset seizures but not all later-onset (31). Our study did not directly compare the seizure onset between the mouse model and patients with PDE. Nonetheless, the studies from different inbred mouse models of PDE suggest that the correlation between clinical phenotypes and the genetic background of patients with PDE should not be overlooked.

Although the high dose of pyridoxine treatment could confer seizure control, at least 75% of patients with PDE suffer from intellectual disability notwithstanding this intervention (1). It points to unique mechanisms underlying the intellectual disability in PDE. By analyzing our PDE mouse model, we found that ALDH7A1-deficient mice with seizure control performed poorly in the cognitive tests, particularly the hippocampus-dependent learning tests, suggesting that ALDH7A1 is necessary, especially for the cognitive functions that require an intact hippocampal function. Neurogenesis persists in the adult mammalian hippocampus throughout life, which plays an important role in cognition and mental health (4–6). Moreover, alteration of adult hippocampal neurogenesis was observed in our PDE mouse model, which is independent of pyridoxine treatment. Specific deletion of ALDH7A1 from adult NSCs sufficiently led to impaired performance on hippocampus-dependent learning tasks, indicating that defective hippocampal neurogenesis may directly contribute to intellectual disability seen in PDE even with seizure control. It is worth noting that mutations in α-aminoadipate semialdehyde synthase, encoding an enzyme upstream of ALDH7A1 in the lysine metabolic pathway, do not affect the proliferation and differentiation of NPCs during embryonic brain development (32), which suggests that impaired saccharopine pathway for lysine metabolism is most likely dispensable for embryonic neurogenesis.

The seizure occurrence was well controlled by pyridoxine treatment; however, our study demonstrated that pyridoxine administration neither remedies the accumulation of toxic lysine metabolites (AASA/P6C) nor ameliorates defective adult hippocampal neurogenesis and cognition in ALDH7A1-deficient mice. It indicates a pyridoxine-independent mechanism underlying the impairment of brain function in PDE. Our transcriptome analyses revealed that pyrimidine metabolism was impaired in ALDH7A1-deficient NSCs, which was not dependent on pyridoxine levels. Moreover, AASA/P6C administration also led to reduced expression of genes involved in de novo pyrimidine biosynthesis. It suggests that ALDH7A1 deficiency leads to AASA/P6C accumulation, which in turn impairs pyrimidine de novo biosynthesis in adult NSCs. We found that supplementation of pyrimidines rescues defective adult hippocampal neurogenesis and cognitive impairment in ALDH7A1-deficient mice. Our findings not only demonstrate that impairment of de novo pyrimidine biosynthesis in ALDH7A1-deficient NSCs leads to altered adult hippocampal neurogenesis, thereby resulting in impaired cognitive functions, but also provide a potential therapeutic intervention to ameliorate the defective mental capacities in PDE patients with seizure control.

Lysine is catabolized in mammals through the pipecolate and saccharopine pathways, in which the former is traditionally believed to be a major degradative pathway for lysine in cerebral tissue (33), and the latter is mainly hepatic and renal (34). However, note that recent studies also offer arguments in favor of the existence of an active saccharopine pathway in the brain (32, 35). By tracking ¹⁵N-labeled lysine at its α- or ε-N atoms, we revealed that ¹⁵N-labeled AAA primarily derives from lysine labeled at the α-amino group resulting from saccharopine pathway activity by ε deamination of lysine in the brain, which is consistent with previous study (16). In support of the notion that the saccharopine pathway is the predominant degradative pathway for lysine in the brain, we further found that incorporation of α-¹⁵N, but not ε-¹⁵N, from lysine into saccharopine and AASA/P6C was significantly accumulated in the brain of ALDH7A1-deficient mice. Loss of ALDH7A1 should inhibit the production of AAA; we anticipated that AAA levels would be decreased by ALDH7A1 deficiency. Unexpectedly, an increase in AAA, a downstream metabolite of lysine catabolism by ALDH7A1, was observed in the brains of Aldh7a1-cKO mice. Moreover, as the precursors for AAA, saccharopine or pipecolic acid treatment did not affect NPC proliferation and differentiation. These data suggest that an alternate route of AAA production from lysine may exist in the brain independent of saccharopine and pipecolate pathways (17, 18).

MATERIALS AND METHODS

Animal

Mice were housed in the animal facility at the Institute of Genetics and Developmental Biology (IGDB), Chinese Academy of Sciences, on a 12-hour reverse light/dark cycle with lights on at 0800 hours. All procedures and husbandry were performed according to protocols approved by the Institutional Animal Care and Use Committee at IGDB. All mice in the study were backcrossed to the C57BL/6 N background for at least six generations. The Aldh7a1^f/f mice were created by Biocytogen. Aldh7a1-cKO mouse line was generated by crossing with Aldh7a1^f/f mice with Nestin-Cre driver mouse (36). Aldh7a1-iKO mouse line was generated by crossing Nestin-CreER^T2 (JAX, 016261) (37) and Rosa26-stop-tdTomato reporter mice (38).

Thymidine analog administration

For analysis of cell proliferation, mice were injected with EdU [200 mg/kg body weight, ip (intraperitoneal) injection, Sigma-Aldrich, E9386] and analyzed 2 hours or 30 days later. For analysis of cell cycle exit, mice were injected with EdU (200 mg/kg body weight, ip injection) and analyzed 24 hours later.

Tamoxifen administration

To induce recombination, mice received tamoxifen (Sigma-Aldrich, T5648) daily for 3 days (180 mg/kg body weight, ip injection, 30 mg/ml in 10% ethanol/sunflower oil, Sigma-Aldrich) based on a published procedure (39).

α-¹⁵N-Lysine and ε-¹⁵N-lysine administration

Mice were fasted overnight and intraperitoneal injected with lysine solutions containing 20 mg of l-[α-15 N]lysine (Cambridge Isotope Laboratories, NLM-143-0) or l-[ε-15 N]lysine (Cambridge Isotope Laboratories, NLM-631-0).

Chemical synthesis of P6C/AASA

The synthesis of P6C/AASA was performed as described previously (3). Allysine ethylene acetal (10 mg; Sigma-Aldrich, 714208) was deblocked by mixing with Amberlyst 15 ion exchange resin (40 mg; Sigma-Aldrich, 216380) in 1 ml of water. The Amberlyst 15 was then used to prepare a solid-phase column and was washed twice with 1 ml of water. P6C/AASA was eluted from the resin with 1.5 ml of 25% ammonia solution (RHAWN, R051454) and was dried under nitrogen at room temperature before resuspending in 1 ml of water.

EEG recording

For electrode placement, mice were anesthetized and placed in a stereotaxic frame. After exposing the cranium, three holes were drilled over the mice cortices (bregma coordinates were AP = +1.2 mm, ML = +1.8 mm; AP = −2.3 mm, ML = ±1.7 mm). Four electrodes (DSI, USA) divided into two channels and implanted in the scull were used to record EEG signals. The radio transmitter (HD-X02) was implanted subcutaneously under the back. After 14 days of recovery from surgery, EEG recording was performed in freely moving animals using an RPC-1 receiver (DSI) with Penomah software. EEG data were analyzed with NeuroScore software. Spontaneous seizures were identified by spiking activity that persisted for at least 15 s with spike amplitudes greater than 2× background amplitude.

Isolation, culture, and in vitro analyses of NPCs

NPCs used in this study were isolated from the DG of 8 to 10-week-old male Aldh7a1^f/f mice based on the published method (40). The cells were maintained in DMEM/F-12 medium containing basic fibroblast growth factor (FGF-2; 20 ng/ml; PeproTech, K1606), epidermal growth factor (20 ng/ml; PeproTech, A2306), 1% B27 supplement (GIBCO, 17504-044), 1% Antibiotic-Antimycotic (GIBCO, 15240062), and 2 mM l-glutamine (GIBCO, 25030081) in a 5% CO₂ incubator at 37°C. Half of the medium was replaced every 2 days.

Proliferation analyses were performed as described previously (38). NPCs were dissociated with trypsin and plated on poly-l-ornithin/laminin-coated slides at a density of 50,000 cells per well in a proliferation medium. The EdU (5 μM) was added into the culture medium for 6 to 8 hours. NPCs were then washed with tris-buffered saline (TBS) and fixed with 4% paraformaldehyde for 30 min at room temperature. To detect EdU incorporation, fixed cells were performed immunocytochemistry analyses.

Differentiation analyses were performed as described previously (38). NPCs were cultured in differentiation medium, DMEM/F12 (1:1), containing 5 μM forskolin (Sigma-Aldrich, #F-6886) and 1 μM retinoic acid (Sigma-Aldrich, #R-2625) for 3 days, followed by fixation with 4% paraformaldehyde for 30 min, washed with phosphate-buffered saline (PBS) for 30 min, and then followed immunocytochemistry analyses.

For in vitro treatment of aNSCs, AASA/P6C, saccharopine (Sigma-Aldrich, S1634-250MG), pipecolic acid (MedChemExpress, HY-Y0669), AAA (APExBIO, 5474), pyridoxine (Sigma-Aldrich, P9755), pyrimidines (uridine: Aladdin, U108811-25g; thymidine: HARVEYBIO, NL1133; Cytidine: HARVEYBIO, NL1136), or leflunomide (Aladdin, L129518) was added to NPCs, then followed by proliferation and differentiation assay.

Isotopic labeling was carried out as described previously (13). Briefly, ¹³C-glucose (Cambridge Isotope Laboratories, CLM-1396-1) or ¹³C-glutamine (Cambridge Isotope Laboratories, CLM-1822-H-0.1) was added to NPCs in a proliferating medium for 22 hours, and then followed by gas chromatography–mass spectrometry (GC-MS) analysis.

Immunocytochemistry analysis

Immunocytochemistry staining was carried out as described previously (38). Briefly, NPCs were pre-blocked using TBS containing 5% normal goat serum (VECTOR, S-1000) and 0.1% Triton X-100 for 30 min, followed by overnight incubation with primary antibodies: anti-Tuj1 (1:1000; Promega, G7121) and rabbit anti-cleaved caspase-3 (1:500; Cell Signaling Technology, 9664). After washing with Dulbecco's PBS (DPBS), cells were incubated with donkey anti-mouse Alexa Fluor 488 (1:1000; Invitrogen, A21202), donkey anti-rabbit Alexa Fluor 568 (1:1000; Invitrogen, A10042), followed by counterstaining with the fluorescent nuclear dye, 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, B2261). The coverslips were mounted with polyvinyl alcohol mounting medium with DABCO (Sigma-Aldrich, 10981).

The cells were quantified using a Nikon-ECLIPSE 80i microscope with NIS-Elements, BR. 3.00 software. The percentage of EdU⁺ cells was calculated as the number of marker-positive cells divided by the total number of DAPI-positive cells.

Tissue preparation and immunohistochemistry

Immunohistological analysis of mouse brains was performed as described previously (41, 42). Mice were euthanized by intraperitoneal injection of Avertin and then transcardially perfused with saline followed by 4% paraformaldehyde (PFA). Brains were dissected out, postfixed overnight in 4% PFA, and then equilibrated in 30% sucrose. Forty-micrometer brain sections were generated using a sliding microtone and stored in a −20°C freezer as floating sections in 96-well plates filled with cryoprotectant solution [glycerol, ethylene glycol, and 0.1 M phosphate buffer (pH 7.4), 1:1:2 by volume].

The brain sections were pre-blocked with TBS++ (TBS containing 3% goat or donkey serum and 0.3% Triton X-100) for 1 hour at room temperature, followed by incubation with primary antibodies diluted in TBS++ overnight at 4°C. After washing three times, secondary antibodies were incubated for 1 hour at room temperature. All sections were counterstained with DAPI.

The primary antibodies were used as follows: anti-ALDH7A1 (1:100; Abcam, ab53278), rabbit anti-Ki67 (1:1000; Abcam, ab15580), mouse anti-Sox2 (1:1000; Abcam, ab79351), rabbit anti-Sox2 (1:1000; Abcam, ab97959), goat anti-GFAP (1:1000; Abcam, ab53554), rat anti-Tbr2 (1:1000; Invitrogen, 14-4875-82), mouse anti-NeuN (1:500; Abcam, ab104224), rabbit anti-DCX (1:1000; Cell Signaling Technology, 4604S), and mouse anti-Mcm2 (1:1000; BD Biosciences, 610701). Fluorescent secondary antibodies were used: donkey anti-chicken 488 (1:1000; Jackson ImmunoResearch, 703-545-155), donkey anti-goat 568 (1:1000; Abcam, ab175474), donkey anti-rabbit 647 (1:1000; Invitrogen, A31573), donkey anti-mouse 647 (1:1000; Invitrogen, A31571). After staining, sections were mounted, coverslipped, and maintained at 4°C in the dark until analysis.

EdU analysis

EdU staining was followed the manufacture protocol (Beyotime, C0081L).

Quantification and fate mapping of tdT⁺ cells in the brain

Quantification and fate mapping of tdT⁺ cells in the brain were performed as described previously (43). For quantification of the phenotype of tdT⁺ cells (double-labeled with either GFAP, Tbr2, DCX, or NeuN), 1 in 12 serial sections starting at the beginning of the hippocampus (relative to bregma, −1.5 mm) to the end of the hippocampus (relative to bregma, −3.5 mm) were used. The cells with radial glial-like morphology were only included in NSC counts. All indicated cells in the brain sections (four to six sections per mouse) were counted inside the section center between 5-μm guard zones of the section surfaces under a Nikon-ECLIPSE 80i microscope with unbiased stereology (StereoInvestigator, MBF Biosciences Inc). The data were presented as a number of cells in a cubic millimeter of DG.

Isotopic labeling analysis

Isotopic labeling of extracted intracellular metabolites was determined by GC-MS. The dried intracellular metabolites were first dissolved in 50 μl of 2% methoxylamine hydrochloride in pyridine and incubated at 37°C for 90 min on a heating block. Next, 80 μl of N-(tert-Butyldimethylsilyl)-N-methyl-trifluoroacetamide with 1% tert-butyldimethylchlorosilane (Thermo Fisher Scientific) was added and the samples were incubated for 30 min at 60°C. After overnight incubation at room temperature, the derivatized samples were briefly centrifuged and the clear liquid was transferred into GC vials for GC-MS analysis. GC-MS analysis was performed on an Agilent 7890B GC system equipped with a DB-5MS capillary column (30 m, 0.25 mm i.d., 0.25-μm phase thickness; Agilent Scientific), connected to an Agilent 5977A mass spectrometer operating under ionization by electron impact at 70 eV.

RNA-seq analysis

Extraction of total RNA was used TRIzol (Invitrogen, 15596018) based on the manufacturer’s protocol. The integrity of the extracted total RNA was analyzed using Agilent 2100 Bioanalyzer (Agilent Technologies). The samples with RNA Integrity Number (RIN) ≥ 7 were subjected to the subsequent analysis. The libraries were constructed using TruSeq Stranded mRNA Sample Prep Kit (Illumina) according to the manufacturer’s instructions. The standard Illumina protocol was used to prepare the libraries for RNA-seq. RNA-seq performed using Illumina HiSeq2500 > 45 million 2 × 100 reads per sample was produced and alignment was performed using mouse genome database GRCm38 version 67. Only uniquely and properly mapped read pairs were used for further analysis. FPKM (fragments per kilobase of transcript per million fragments mapped) value of each gene was calculated using cufflinks, and the read counts of each gene were obtained by htseq-count. DEGs were identified using the DESeq (2012) functions “estimate Size Factors” and “nbinom Test.” P < 0.05 and fold change > 0.2 were set as the threshold for significantly differential expression. Hierarchical cluster analysis of DEGs was performed to explore gene expression patterns. Gene Ontology enrichment and KEGG pathway enrichment analysis of DEGs were respectively performed using R based on the hypergeometric distribution.

RNA isolation and real-time PCR

RNA isolation was used TRIzol (Invitrogen, 15596018) based on the manufacturer’s protocol. The first-strand cDNA was generated by reverse transcription with oligo(dT) primer (Promega, A5001). Standard RT-PCR was performed using GoTaq DNA polymerase (Promega, M3001). To quantify the mRNA levels using real-time PCR, aliquots of first-stranded cDNA were amplified with gene-specific primers and SYBR Green PCR Master Mix (CWBIOTECH, CW0682A) using a Bio-Rad Real-Time PCR System (CFX96). The sequences of primers used for PCR reactions are listed in table S4.

Osmotic pump grafting

Micro-osmotic pump grafting was performed as described previously (44). The osmotic pump was prepared according to the manufacturer’s procedure (RWD, 1001 W, flow rate of 0.5 μl per hour for 7 days or 1004 W, flow rate of 0.125 μl per hour for 28 days). Micro-osmotic pumps were filled with pyrimidines (uridine: Aladdin, U108811-25 g; thymidine: HARVEYBIO, NL1133; cytidine: HARVEYBIO, NL1136) or DPBS, and then positioned at the following coordinates relative to bregma, caudal: −2.0 mm; lateral: ±1.7 mm; ventral: −1.9 mm.

Recombinant lentivirus production

Lentivirus production was performed as described previously (39). Briefly, lentiviral transfer vector DNA and packaging plasmid DNA were cotransfected into 293 T cells (American Type Culture Collection, #CRL-3126). The medium was collected and pooled at 40, 64, and 88 hours, and then filtered through a 0.22-μm filter. Viruses were concentrated by ultracentrifuge at 19,000 rpm for 2 hours at 4°C using a SW32 rotor (Beckman). Viruses were washed once with PBS and then resuspended in 150 μl of PBS. We routinely obtained 1 × 10⁹ infectious viral particles per milliliter and a ~80% transduction rate in cultured NPCs.

Western blot assay

The NPCs or brain tissues were lysed in radioimmunoprecipitation assay buffer [20 mM tris-HCl (pH 7.5), 100 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, and 1 mM phenylmethylsulfonyl fluoride) containing a Complete Protease Inhibitor Cocktail (Roche, 11697498001). Cell lysates were centrifuged at 12,000 rpm for 10 min. Fifty micrograms of supernatants was resolved on SDS–polyacrylamide gel electrophoresis and blotted with the indicated antibodies. Primary antibodies used were the following: anti-ALDH7A1 (1:1000; Abcam, ab53278), anti-pCHK1 (1:1000; Abclonal, A21009), anti-CHK1 (1:1000; Abclonal, A7653), and mouse anti-GAPDH (1:50,000; Proteintech, 60004-1-Ig). The amount of GAPDH was used as the loading control.

Novel object recognition test

Mouse was habituated to an empty white chamber by allowing them to freely explore for 15 min. After 24 hours, every mouse was rehabituated to the empty chamber for 1 min and then placed in a holding cage while two identical objects (metallic cones with 9 cm height) were placed in the corners of the arena 7 cm from the walls. Mice were returned to the chamber for training and allowed to freely explore until they accumulated a total of 30 s exploring the objects (exploration recorded when the front paws or nose contacted the object). Mice were then removed from the chamber, immediately infused, and returned to their home cage. After 24 hours, object recognition was tested by substituting novel objects (dolls with 5-inch height). Time spent with each object was recorded. The time spent exploring each object for 10 min in the novel object tests by watching recorded behavior (via digital video) was measured. Scoring can be done by using two stopwatches and collecting exploratory time for each object simultaneously. Novel object preference is expressed as a percentage of time exploring the novel location among the cumulative time spent exploring both objects.

Novel object location test

Mouse was habituated to an empty white chamber by allowing them to freely explore for 15 min (four mice per time point). After 24 hours, every mouse was rehabituated to the empty chamber for 1 min and then placed in a holding cage, while two identical objects were placed in the corners of the arena 7 cm from the walls. Mice were returned to the chamber for training and allowed to freely explore until they accumulated a total of 30 s exploring the objects (exploration recorded when the front paws or nose contacted the object). Mice were then removed from the chamber, immediately infused, and returned to their home cage. After 24, object recognition was tested by substituting NOL counterbalanced across mice. Time spent with each object was recorded. The time spent exploring each object for 10 min in the novel location tests by watching recorded behavior (via digital video) was measured. Scoring can be done by using two stopwatches and collecting exploratory time for each object simultaneously. Novel location preference is expressed as a percentage of time exploring the novel location among the cumulative time spent exploring both objects.

Fear conditioning test

Mouse was placed into a shock chamber and allowed to explore for 2 min. Then, a white noise tone (87 dB) sounded for 30 s (conditional stimulus or “CS”). During the last 1.5 s of the tone, mice received a mild footshock (0.5 mA) (unconditioned stimulus or “US”). The same tone-footshock (CS-US) combination was delivered again 2 min later. This cycle was presented a total of three times with a 60-s interval. The context test was performed 24 hours after the training. During the test, mice were placed back into the same training chamber, and monitored by an overhead camera in the chamber for 5 min. The cue test was performed 2 hours after the context test, in which colored plexiglass inserts were placed into the training chamber to hide the shock grid and to change the “context” of the chamber. Mice were then placed in the chamber and monitored by the overhead camera for 6 min, during which two CS (spaced the same way as in the training session) were given. The events in the fear conditioning test were programmed and data recorded through the Startle and Fear conditioning system (Panlab) and Packwin software (V2.0.05).

Y-Maze test

Mice were placed into one of the arms of the maze (start arm) and allowed to explore the maze with one of the arms closed for 15 min (training trial). After a 1-hour intertrial interval, mice were returned to the Y maze by placing them in the start arm. Then, the mice were allowed to explore freely all three arms of the maze for 5 min (test trial). The percentage of spontaneous alternation was calculated as [number of spontaneous alternations/(number of total arm entries − 2)] × 100. All mouse behavior was recorded by the Limelight 5 software.

Open-field assay

Mouse was placed in an unfamiliar arena with clear side walls (10 inch × 10 inch × 16 inch; RWD life Science) and were allowed to freely explore the arena for 20 min. They were returned to their home cages after the test. Their locomotor activity was tracked by photo beams preinstalled to the arena and then analyzed by Panlab SMART 3.0 Software.

Elevated plus maze test

Mouse was placed in an elevated plus maze for 5 min and tracked using Panlab SMART 3.0 Software. The amount of time spent in the open arms was measured.

Statistical analysis

All experiments were performed and analyzed by the same experimenter, blind to the animals’ genotype or group treatment under assessment. No statistical methods were used to predetermine sample size in other experiments. Sample sizes can be found within figure legends. All percentages were arcsine-transformed before statistical analysis. All data were shown as means with standard error of mean (means ± SEM). Statistical analysis was performed using one-way analysis of variance (ANOVA), two-way ANOVA, and two-tailed Student’s t test with the aid of SPSS version 22 and GraphPad software. The D’Agostino and Pearson normality test was performed to verify normality. For the one-sample t test, we first normalized the treatment group by the control group, and then one-sample t test against mean of 1 was used on the normalized values. Probabilities of P < 0.05 were considered significant. For the in vitro analysis, the number of independent experiments was specified in the legend of each figure and each data point was shown as the average per experiment.

Supplementary Materials

This PDF file includes:

Figs. S1 to S11

Legends for tables S1 to S4

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S4