Lamin B1 decline underlies age-related loss of adult hippocampal neurogenesis

Abstract

Neurogenesis in the adult hippocampus declines with age, a process that has been implicated in cognitive and emotional impairments. However, the mechanisms underlying this decline have remained elusive. Here, we show that the age-dependent downregulation of lamin B1, one of the nuclear lamins in adult neural stem/progenitor cells (ANSPCs), underlies age-related alterations in adult hippocampal neurogenesis. Our results indicate that higher levels of lamin B1 in ANSPCs safeguard against premature differentiation and regulate the maintenance of ANSPCs. However, the level of lamin B1 in ANSPCs declines during aging. Precocious loss of lamin B1 in ANSPCs transiently promotes neurogenesis but eventually depletes it. Furthermore, the reduction of lamin B1 in ANSPCs recapitulates age-related anxiety-like behavior in mice. Our results indicate that the decline in lamin B1 underlies stem cell aging and impacts the homeostasis of adult neurogenesis and mood regulation.

Keywords: adult hippocampal neurogenesis; lamin B1; mood regulation; stem cell aging.

Figure 1. Enrichment of lamin B1 in ANSPCs

1. Enrichment of lamin B1 in the SGZ of DG.

2. Immunohistochemical analyses of lamin B1‐high cells. The lamin B1‐high cells express Sox2 or PSA‐NCAM, but not Prox1. Arrowheads indicate Prox1‐negative lamin B1‐high cells in the SGZ.

3. Immunohistochemical analyses of lamin B1 in pNestin‐GFP mice. An open arrowhead indicates RGL‐ANSC, and a closed arrowhead indicates a non‐RGL(NR)‐NPC.

4. A schema for the changes in lamin B1 levels during adult hippocampal neurogenesis.

5. Immunocytochemistry of lamin B1 in NPCs and Tuj1+ differentiating neurons 4 days after the induction of differentiation.

6. Relative expression of Lmnb1 and Tubb3 revealed by qRT–PCR from NPCs and differentiating neurons at the indicated days. ***P < 0.001, **P < 0.01, *P < 0.05, one‐sample t‐test (n = 4).

7. Quantification of signal intensity of lamin B1 immunoreactivity in different neural cell types. F _3,82 = 173.1, P = 1.2 e^‐34, n = 12–30 cells per cell type, one‐way ANOVA followed by Tukey–Kramer test (**P < 0.01).

Data information: A.U. = Arbitrary units. Bar graphs show mean ± SEM. Scale bars, 75 µm in (A), 25 µm in (B lower, C), 10 µm in (B upper, E).

Figure 2. Age‐dependent reduction in lamin B1 in ANSPCs

1. A

   Age‐dependent reduction of lamin B1 in the SGZ. As early as 5.5 months of age, the levels of lamin B1 in the SGZ were markedly reduced. Insets show higher magnification images in the SGZ.

2. B–D

   Quantification of lamin B1 levels using immunofluorescent signals. Lamin B1 levels in Sox2‐positive, PSA‐NCAM‐negative‐ANSPCs were selectively downregulated at 5.5 months of age (ANPCs, P = 1.02129e‐19; Neuroblasts, P = 0.09, DGCs, P = 6.4011^e‐15, ANOVA). **P < 0.001, ANOVA followed by Tukey–Kramer test. (3 animals, 30 cells per conditions)

3. E

   Age‐dependent reduction of lamin B1 in RGL‐ANSCs (upper) and NR‐ANPC (lower) from 2 to 5.5 months of age. Arrowheads indicate RGL‐ANSCs or NR‐ANPCs.

4. F

   Quantification of lamin B1 levels in RGL‐ANSCs and ANPCs. *P < 0.05, ***P < 0.001, (4 animals, 18‐32 cells per conditions, Mann–Whitney test).

5. G

   A hypothetical model. Aging induces the reduction of lamin B1 in ANSPCs, which underlies stem cell aging.

Data information: Scale bars, 100 µm in (A), 10 µm in (E). Bar graphs show mean ± SEM. A.U. = Arbitrary units.

Figure EV1. Generation and evaluation of Lmnb1 conditional knockout mouse line

1. A

   A schematic illustration of Lmnb1 conditional knockout mouse line (cKO).

2. B

   Induction of EYFP in ANSCPs 10 days after the administration of TAM in the SGZ of DG.

3. C, D

   Representative images of lamin B1 complete‐loss cell (arrowhead) and non‐EYFP ANPC (open arrowhead) (C), lamin B1 partial‐loss cell (arrowhead), and non‐EYFP ANPC (open arrowhead) (D).

4. E

   The fractions of EYFP+ cells with lamin B1 deficiency for each cell type 10 days or 3 weeks after TAM administration (Total number of counted cells, 10 days; RGL‐ANSCs. 81 cells, ANPCs = 116 cells, Neuroblasts = 74 cells from 3 animals; 3 weeks, RGL‐ANSCs = 39 cells, ANPCs = 59 cells, neuroblasts = 84 cells from 3 animals). Lamin B1‐deficient cells are defined as having 30% or more reduced levels of lamin B1 intensity compared to the averaged intensity of the same type of control cells.

5. F

   A confocal image of lamin B1 deficiency in Ki67+ proliferating cells (arrows) 10 days after TAM infusion. Open arrows indicate lamin B1 deficient cells without Ki67.

6. G

   The fraction of lamin B1‐deficient cells in Ki67+ proliferating cells. Ki67+ non‐RGL cells (others cells) show lamin B1 deficiency whereas only a few Ki67+ RGL‐ANSCs exhibit laminB1 deficiency, implying the necessity for cellular proliferation to deplete lamin B1 proteins (Total number of counted cells, 10 days; RGL‐ANSCs. 115 cells, non‐RGL cells = 196 cells, from 3 animals; 3 weeks; RGL‐ANSCs. 42 cells, non‐RGL cells = 142 cells, from 3 animals).

Data information: Scale bars, 200 µm in (B, Left), 50 µm in (B, Right), 10 µm (C, D, F).

Figure 3. Precocious reduction of lamin B1 in ANSPCs induces anxiety‐like behaviors

1. A

   Schematic of TAM treatment and behavioral tests.

2. B–F

   Open‐field (OF) test. (B) Color‐coded distribution of the exploratory path in the open‐field test. X‐Y axes of squares correspond to the OF arena. Old WT mice and Lmnb1‐cKO mice spend more time in the periphery of the OF arena. (C) Total distance travelled during the OF test. No significant difference among groups. F _2,25 = 2.23, P = 0.13, ANOVA. (D) Velocity of travel during the OF test. F _2,25 = 1.60, P = 0.22, ANOVA. N.S. (not significant). (E) Fraction of time spent in the periphery. F _{2, 25} = 10.71, P = 0.00044, ANOVA followed by Tukey–Kramer, **P < 0.01. (F) Fraction of travelled distance in the periphery, F _{2, 25} = 3.84, P = 0.035, ANOVA followed by Tukey–Kramer, *P < 0.05 (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

3. G

   Principal components analysis of behavior from the OF test (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test). Circles indicate 66% confidence levels for each group.

4. H–J

   Novelty‐suppressed feeding. (H) Total feeding time during the NSF test, P = 0.0068, Kruskal–Wallis test followed by Dunn’s test. **P < 0.01. (n: WT = 6, WT‐old = 7, Lmnb1‐cKO = 6 for novelty‐suppressed feeding). (I) Food consumption during the NSF test, P = 0.014, Kruskal–Wallis test followed by Dunn’s test. *P < 0.05. (J) Total number of feeding instances during NSF test. N.S. (not significant).

Data information: Bar graphs show mean ± SEM.

Figure EV2. Increased age‐related behavior in Lmnb1‐cKO mice

1. Rearing counts during the OF test. F _2,25 = 3.95, *P = 0.032, ANOVA followed by Tukey–Kramer (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

2. Average time spent on stereotypic behavior. F _{2, 25)} = 7.15, P = 0.0035, ANOVA followed by Tukey–Kramer, *P < 0.05, **P < 0.01 (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

3. Total resting time in the OF. F_{( 2, 25)} = 4.65, P = 0.018, ANOVA followed by Tukey–Kramer, *P < 0.05 (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

4. Total resting time in the peripheral area. F _2,25 = 12.34, P = 0.00019, ANOVA followed by Tukey–Kramer, **P < 0.01, ***P < 0.001 (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

5. Ratio of resting time between center and peripheral area. Kruskal–Wallis test followed by Dunn test, *P < 0.015, **P < 0.01 (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

6. No significant changes in spontaneous alternation in Y‐maze test. F _2,25 = 0.12, P = 0.89, ANOVA (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11).

7. Preference to novel object was not different among three groups F _2,25 = 3.39, P = 0.45, ANOVA) (n: WT = 10, WT‐old = 7, Lmnb1‐cKO = 11 for the OF test).

8. Proportional body weight loss after 24‐h food deprivation for the NSF test F _2,25 = 6.11, P = 0.94, ANOVA) (n: WT = 6, WT‐old = 7, Lmnb1‐cKO = 6).

9. Latency to feed in NSF test. Kruskal–Wallis test followed by Steel test, *P = 0.038, (n: WT = 6, WT‐old = 7, Lmnb1‐cKO = 6).

Data information: Data are presented as mean ± SEM.

Figure 4. Lamin B1‐knockout promotes differentiation of ANPCs in the short term

1. A

   Schematic of TAM treatment and collection of brain tissue.

2. B, C

   Representative images of Control and cKO with EYFP and DCX staining 3 weeks after TAM treatment. Note that the numbers of DCX‐positive cells are increased. Boxes indicate magnified regions in (C).

3. D–F

   Quantification 3 weeks after TAM treatment. Among EYFP‐positive cells, the density of RGL‐ANSCs is not different (P = 0.65, t‐test) (D). The density of ANPCs is significantly decreased (P = 0.0037, t‐test) (E) whereas the density of DCX‐positive cells is markedly increased (P = 0.014, t‐test) (F). (n = 4 for Control, n = 3 for cKO). Open arrowheads indicate respective cells.

4. G–I

   Quantification two months after the induction of KO (n = 3 for Control, n = 4 for cKO). The density of RGL‐ANSCs is not significantly different (P = 0.29, t‐test) (G). The density of ANPCs is significantly decreased. (P = 0.0092, t‐test) (H). The density of neuroblasts is not significantly different (P = 0.76, t‐test) (I).

5. J–L

   Quantification of proliferating cells 3 weeks after TAM treatment. The density of Ki67+ RGL‐ANSCs is significantly decreased (P = 0.027, Mann–Whitney test) (J) and that of Ki67+ ANPCs is also significantly decreased. (P = 0.034, t‐test) (K). The density of Ki67+ neuroblasts is significantly increased (P = 0.0011, t‐test) (N) (n = 4 for Control, n = 4 for cKO) (L).

6. M–O

   Quantification of proliferating cells two months after TAM treatment. The density of Ki67+ RGL‐ANSCs is significantly decreased (P = 0.0086, t‐test) (M), and densities of Ki67+ ANPCs and Ki67+ neuroblasts are also significantly decreased. (P = 0.023, P = 0.025, t‐test) (n = 3 for Control, n = 3 for cKO) (N, O).

7. P

   Declined return to RGL‐ANSCs. Left, experimental scheme. Analysis was performed 3 weeks after BrdU injections. Middle, a representative image of BrdU+ GFAP+ RGL‐ANSCs. Right, Quantification of BrdU+ GFAP+RGL‐ANSCs (P = 0.019, n = 4 for Control, n = 4 for cKO).

Data information: Data presented as mean ± SEM. Scale bars, 100 µm in (B), 25 µm in (C, P) and 10 µm in (D, F, H). *P < 0.05, **P < 0.01, ***P < 0.001, t‐test. Source data are available online for this figure.

Figure EV3. Effects of lamin B1 loss on adult hippocampal neurogenesis

1. A

   Schematic of BrdU treatment and collection of brain tissue.

2. B

   Representative images of BrdU staining in the DG 3 weeks after TAM administration. BrdU‐positive cells were markedly increased in cKO mice.

3. C

   Quantification of BrdU + cells 3 weeks after TAM administration (*P = 0.021, t‐test, n = 4).

4. D

   Quantification of BrdU + cells 2 months after TAM administration (***P = 0.0007, t‐test, n = 3 for Control, 4 for cKO).

5. E, F

   Orthogonal views of confocal images to identify cell type of EFYP + cells.

6. G

   Schematic of EdU treatment and collection of brain tissue.

7. H

   Representative images of EdU‐click staining in the DG. EdU‐positive cells were markedly reduced in cKO mice.

8. I

   Quantification of EdU‐positive cells in the DG indicates that lamin B1 loss induces the prominent reduction of new cell generation in the DG similar to Cont‐15M mice. **P < 0.01, ANOVA (P = 0.0009) followed by Tukey–Kramer test (n = 4 for each genotype).

Data information: Data are presented as mean ± SD. Scale bars, 100 µm in (B), 50 µm in (E, F), and 200 µm in (H).

Figure 5. Lamin B1‐knockout depletes adult neurogenesis in the long term

1. A

   Schematic of TAM treatment and collection of brain tissue.

2. B, C

   Representative confocal images of the SGZ 6.5 months after the treatment with TAM.

3. D–I

   Quantification of cell numbers 6.5 months after the induction of KO (n = 4 for Cont‐8M, n = 3 for cKO‐8M, n = 4 for Cont‐15M,). (D) The density of RGL‐ANSCs tends to be reduced in cKO. F _2,8 = 3.98, P = 0.063, one‐way ANOVA. (E) The density of ANPCs was reduced in cKO‐8M and Cont‐15M mice. F _2,8 = 11.49, P = 0.0044, one‐way ANOVA followed by Tukey’s HSD test (**P < 0.01). (F) The density of DCX‐positive cells was reduced in WT‐old and cKO mice. P = 0.0091, Kruskal–Wallis test followed by Dunn’s test (*P < 0.05). (G) The density of Ki67+ RGL‐ANSCs was reduced in cKO‐8M mice (*P = 0.029, Kruskal‐Wallis test followed by Dunn’s test). (H) The density of Ki67+ ANPC was not significantly different among three groups (P = 0.49, Kruskal–Wallis test). (I) The density of NeuN+ EYFP+ neurons was reduced in cKO‐8M mice. F _2,8 = 34.85, P = 0.00011, one‐way ANOVA followed by Tukey’s HSD test (**P < 0.001). (n = 4 for Cont‐8M, n = 3 for cKO‐8M, n = 4 for Cont‐15M).

Data information: Data represent mean ± SEM. Scale bars, 100 µm in (B), 50 µm in (C).

Figure EV4. Lamin B1 is essential for the survival of adult‐born cells

1. A

   Quantification of EYFP+ NeuN+ neurons 2 months after TAM administration. *P = 0.042, t‐test, n = 4).

2. B

   Schematic of BrdU treatment and collection of brain tissue.

3. C

   Representative images of BrdU staining 10 days or 2 months after TAM treatment. Scale bar = 250 µm.

4. D

   Quantification of BrdU‐positive cells in each time point. Number of BrdU‐positive cells was significantly higher at day 10 (*P = 0.049, t‐test) or 3 weeks (*P = 0.019, t‐test) after TAM in cKO mice, but not at 2 months (P = 0.12, t‐test, n = 3 animals). Data represent mean ± SD.

5. E

   The survival ratio of BrdU + cells from 10 days to 2 months after TAM treatment (*P = 0.015, t‐test, n = 3). Data represent mean ± SD.

6. F

   Representative images of active caspase3 2 months after TAM treatment. Scale bar = 20 µm.

7. G–I

   Significant difference in the density of active caspase3 was observed 2 months after TAM treatment, but not 3 weeks or 6.5 months after TAM treatment (3 weeks, P = 0.86, n = 3, t‐test; 2 months, *P = 0.028, n = 5‐6, t‐test; 6.5 months, P = 0.09, n = 4, ANOVA). Data are presented as mean ± SD.

8. J

   Experimental schema. RV:pSox2‐Cre‐GFP was injected into the DG of Lmnb1fl/fl:Ai3 or Lmnb1+/+:Ai3 mice and the morphology of EYFP+ newborn neurons was assessed at 21dpi. Scale bar = 20 µm.

9. K

   Dendritic reconstruction of RV‐labeled neurons derived from WT and cKO mice. Scale bars = 20 µm.

10. L, M

    Total dendrite length (L) and total numbers of branches (M) were quantified (65 cells for control, 35 cells for cKO from 3 animals per genotype; total length, *P = 0.0147; number of branches, P = 0.099, Mann–Whitney test). Data are presented as mean ± SEM.

11. N

    Sholl analysis of neurite length. **P < 0.01, *P < 0.05 (t‐test) (65 cells for control, 35 cells for cKO from 3 animals per genotype). Data are presented as mean ± SEM.

Figure 6. Lamin B1‐dependent gene regulation in NPCs

1. A

   Schematic of Lmnb1‐KO in NPCs. After the introduction of LV, NPCs were kept in proliferative conditions with FGF2 and EGF.

2. B

   MA plot of differentially expressed genes between control and Lmnb1‐KO NPCs.

3. C, D

   Gene ontology analyses in LaminB1‐regulated genes.

4. E

   Fraction of direction of gene regulation by LaminB1. Genes were allocated dependent on where LaminB1 interacts with genes. Red bars indicated upregulated genes, and blue bars indicate downregulated genes after lamin B1 knockout.

5. F

   Examples of lamin B1‐directed genes in NPCs. Blue bars indicate laminB1‐LAD.

Figure EV5. Gene pathways regulated by lamin B1 in NPCs

1. A

   Confirmation of lamin B1 depletion after the delivery of LV; pSox2‐Cre. Closed white arrowheads indicate the depletion of lamin B1 immunoreactivity in right panels. Scale bar = 25 µm.

2. B, C

   GO and KEGG pathway enrichment analyses of differentially expressed genes.

3. D, E

   GO and KEGG enrichment analyses for lamin B1‐directed upregulated genes.

4. F

   Fold change expression of Bmp4 and Id1‐4 from the transcriptomic analysis in cKO‐NPCs, n = 3, DEseq2 for statistical test. Data are presented as mean ± SEM.

5. G

   qRT–PCR validation of Bmp4 and Id4 upregulation in cKO‐NPCs (Bmp4, *P = 0.022; Id4, *P = 0.015, one‐sample t‐test, n = 3). Data are presented as mean ± SD.

Figure 7. Exogenous expression of laminB1 represses neural differentiation

1. A

   Experimental schema. After the introduction of RV into NPCs in a proliferative condition, FGF2 was withdrawn for 2 days, and samples were collected.

2. B, C

   Confirmation of Lamin B1 overexpression by qRT–PCR (*P = 0.020, n = 3, one‐sample t‐test), and immunocytochemistry in NPCs. Scale bar = 10 µm.

3. D

   Relative induced expression of differentiation markers upon the withdrawal of FGF2. 0 % indicates the same expression levels with control. Exogenous expression of lamin B1 inhibits the induction of genes related to neural differentiation (Tuj1, *P = 0.023; Prox1, ***P = 0.00012; NeuroD1, **P = 0.0046; Ascl1, ***P = 0.00064; S100β, ***P = 0.000097; n = 3, one‐sample t‐test).

4. E

   Experimental schema. RV:EGFP‐IRES‐EGFP or LaminB1‐IRES‐EGFP was injected into the DG of wild‐type mice, and the brains were collected 7 days later.

5. F

   Confocal images after exogenous expression of lamin B1 in EGFP+ cells with RV; LaminB1‐IRES‐GFP (laminB1‐OE.) Scale bar = 20 µm.

6. G, H

   Exogenous expression of laminB1 repressed neuronal differentiation in vivo. An arrowhead indicate DCX+ GFP+ cell in control (G, left) and open arrowheads indicate Sox2+ GFP+ cells in lamin B1‐OE cells. The fraction of EGFP+ DCX+ cells was significantly reduced by lamin B1 overexpression (***P < 0.0002, n = 3), whereas the fractions of EGFP+ DCX+ Sox2+ cells and the EGFP+ Sox2+ were increased (EGFP + DCX+ Sox2+, **P = 0.0024; EGFP+ Sox2+, P = 0.058; others, *P = 0.02). Scale bar = 20 µm.

Data information: Data represent mean ± SD.