IKKβ/NF-κB disrupts adult hypothalamic neural stem cells to mediate a neurodegenerative mechanism of dietary obesity and pre-diabetes

Abstract

Adult neural stem cells (NSCs) are known to exist in a few regions of the brain; however, the entity and physiological/disease relevance of adult hypothalamic NSCs (htNSCs) remain unclear. This work shows that adult htNSCs are multipotent and predominantly present in the mediobasal hypothalamus of adult mice. Chronic high-fat-diet feeding led to not only depletion but also neurogenic impairment of htNSCs associated with IKKβ/NF-κB activation. In vitro htNSC models demonstrated that their survival and neurogenesis markedly decreased on IKKβ/NF-κB activation but increased on IKKβ/NF-κB inhibition, mechanistically mediated by IKKβ/NF-κB-controlled apoptosis and Notch signalling. Mouse studies revealed that htNSC-specific IKKβ/NF-κB activation led to depletion and impaired neuronal differentiation of htNSCs, and ultimately the development of obesity and pre-diabetes. In conclusion, adult htNSCs are important for the central regulation of metabolic physiology, and IKKβ/NF-κB-mediated impairment of adult htNSCs is a critical neurodegenerative mechanism for obesity and related diabetes.

Figure 1

In vivo and in vitro definition of htNSCs in adult mice. (a–c) Brain sections across the ARC were made from C57BL/6 mice (a) or Nestin-Cre mice (b,c) for immunostaining. Brain sections across dentate gyrus were included in (a). Mice were on chow-fed and ~3-month-old males. Nuclear staining by DAPI (blue) revealed all cells in the sections. Merged images show the co-distribution of the indicated molecular markers. Chow-fed males at 3 months of age were used in these studies. (c) Numbers of Nestin⁺, Sox2⁺, or Nestin⁺Sox2⁺ cells in the MBH (comprising the ARC and VMH) were compared to numbers of tanycytes in lateral third-ventricle wall (L3V, T) vs. median eminence (ME, T). n = 6 mice (c). Scale bar = 50 μm (a,b). (d,e) Hypothalamic tissues were sampled from normal C57BL/6 mice (chow-fed, 3 months old) for neurosphere culture as described in Method section. Neurospheres were formed and passaged in growth medium containing bFGF and EGF. Neurospheres at various passages were attached to slides and immunostained for Sox2 (d), nestin and Blbp (e). Images were merged with DAPI staining to reveal the nuclear distribution of Sox2 and the cytoplasmic distribution of nestin and Blbp. Scale bar = 50 μm. (f) The hypothalamus and various other brain components were sampled from normal C57BL/6 mice (chow-fed, 3 months old) for the neurosphere (NS) assay. Data shows the total number of primary NS (without passage) normalized by the mass (mg) of brain tissue from which NS were derived. Hy: hypothalamus; Co: cortex; Po: pons; Th: thalamus; Ce: cerebellum; DG: dentate gyrus. ** P < 0.01, *** P < 0.001, n = 4 mice per group; error bars reflect means ± s.e.m. (g) Neurospheres were derived from the hypothalamus of normal mice (chow-fed, 3 months old). Dissociated neurospheric cells at the same passage were induced to differentiate as described in the method. Following 7-day differentiation, cells were immunostained for neuronal marker Tuj1, astrocyte marker GFAP, and oligodendrocyte marker O4. Nuclear staining of DAPI revealed all cells in the slides. Scale bar = 50 μm.

Figure 2

Brdu tracking of adult htNSCs-mediated neurogenesis in mice. (a) C57BL/6 mice (chow-fed males, 4 months old) received a single-day icv injection of Brdu. Brains were fixed at Day 1 vs. 7 and sectioned for Brdu staining. Total numbers of Brdu-labeled cells in serial ARC sections were counted. (b–i) C57BL/6 mice (chow-fed males, 4 months old) received daily icv injections of Brdu consecutively for 7 days. Brains were fixed at Day 10 vs. 30 and then sectioned for Brdu staining (b) or co-immunostaining with indicated markers (c–i). (b) Total numbers of Brdu-labeled cells in serial ARC sections were counted. (e–i) Total numbers of Brdu-labeled cells co-immunostained with NeuN (e), POMC (f), NPY (g), S100B (h), and RIP (i) in serial ARC sections were counted. ** P < 0.01, *** P < 0.001, n = 6 mice (a,b,g,i), n = 4 mice (e,h) and n = 5 mice (f) per group. Error bars reflect means ± s.e.m. Scale bar = 50 μm (c,d).

Figure 3

Fate mapping of adult htNSCs in mice under normal physiology. ROSA-lox-STOP-lox-YFP mice (chow-fed males, 3 months old) were bilaterally injected in the mediobasal hypothalamus with lentiviruses which directed Cre expression under the control of Sox2 promoter. Following indicated days post viral injection, hypothalamus sections were made for tracking neural differentiation of YFP-labeled cells. (a) Schematic of lentiviral vector and genetic mouse model. Lentivirus expressing Cre under the control of Sox2 promoter was termed P_sox2-Cre lentivirus. The control was a lentiviral vector without containing Cre (schematic not shown). (b–i) Co-imaging of YFP (green) (b–d) with immunostaining (red) of Sox2 (b), NeuN (c) or POMC (d) at indicated days post viral injection. Cell nuclear staining (blue) by DAPI revealed all cells in the sections. Bar graphs: YFP-labeled NeuN-positive cells (YFP⁺NeuN⁺) (e); POMC-positive cells (YFP⁺POMC⁺) (f); NPY-positive cells (YFP⁺NPY⁺) (g); S100B-positive cells (YFP⁺S100B⁺) (h) and RIP-Positive cells (YFP⁺RIP⁺) (i) in serial ARC sections were counted. *** P < 0.001, n = 5 mice (e,f), n = 6 mice (g,i) and n = 4 mice (h) per group. Error bars reflect means ± s.e.m. Scale bar = 50 μm (b–d).

Figure 4

Impaired survival and neurogenesis of htNSCs derived from mice dietary obesity. (a–c) Adult male C57BL/6 mice were maintained under chow vs. HFD feeding for 4 months (a,b) or 8 months (c). Hypothalamic sections were immunostained for Sox2 (a) or POMC (image not shown). DAPI nuclear staining revealed all cells in the section. Sox2-immunoreactive (Sox2⁺) cells (b) and POMC neurons (c) were counted in the arcuate nucleus (ARC). (d–i) C57BL/6 mice were maintained under normal chow vs. HFD feeding for 4 months (4M), and the hypothalami of these mice were removed to generate neurospheres for in vitro assays. (d–f) Morphology (d), number (e), and size (f) of primary neurospheres. (g) Primary neurospheres were passaged with the same initial number (10⁴ cells per group), and cell outputs were followed for 5 passages. (h,i) Neurospheric cells at the same passage (representing Passages 5–10) were induced to differentiate for 7 days, cells were fixed and immunostained for Tuj1 (h), and Tuj1-positive (Tuj1⁺) cells were counted in the slides (i). * P < 0.05, ** P < 0.01, *** P < 0.001, comparisons between chow and HFD at indicated points, n = 5 mice (b), n = 6 mice (c) and n = 4 mice (e,f) per group, and n = 4 per group (g,i); error bars reflect means ± s.e.m. Scale bar = 50 μm (a,d,h).

Figure 5

Impaired in vitro proliferation of htNSCs with IKKβ/NF-κB activation. (a) Adult C57BL/6 mice were fed on a chow vs. HFD for 4 months. Hypothalamic neurospheres were generated from these mice, cultured and passaged in vitro. Western blotting was performed for cultured neurospheric cells to measure phosphorylated RelA (pRelA). RelA and β-actin were analyzed as controls. (b,c) Neurospheres were derived from the hypothalamus of chow-fed C57BL/6 mice (3 months old). Dissociated neurospheric were infected with lentiviruses expressing ^CAIKKβ (GFP-conjugated), ^DNIκBα (GFP-conjugated), and GFP. In vitro models of htNSCs with stable transduction of GFP, ^CAIKKβ and ^DNIκBα were established and maintained in the selection medium, named GFP-htNSCs, IKKβ-htNSCs and IκBα-htNSCs (Suppl. Fig. 7), respectively. Cells were analyzed via Western blots for pRelA. Total protein levels of RelA and β-actin were analyzed as controls. (d–h) Dissociated IKKβ-htNSCs, IκBα-htNSCs, and GFP-htNSCs were maintained in the growth medium. (d,e) Cells (Passage 6) were labeled with Brdu and analyzed for Brdu-positive (Brdu⁺) cells. (f) Cell outputs over subsequent 4 passages were analyzed. (g,h) Cells (Passage 6) were subjected to Tunel staining, and Tunel staining-positive cells were counted. (i–m) Neurospheres derived from the hypothalamus of C57BL/6 mice that received a normal chow vs. HFD for 4 months. Dissociated neurospheric cells were infected with lentiviruses expressing ^DNIκBα (GFP-conjugated) or GFP to establish IκBα-htNSCs^HFD, GFP-htNSCs^HFD, and GFP-htNSCs^chow lines. Dissociated cells with the same initial cell numbers from Passage 6 were maintained in the growth medium. (i,j) Cells (Passage 6) were pulse labeled with Brdu, and Brdu-positive (Brdu⁺) cells were analyzed. (k) Cell outputs from the same initial number at Passage 6 were followed for 4 passages. (l,m) Tunel assay was performed for cells (Passage 6, Day 2) and analyzed using Tunel staining. (b,c,e,f,h) IKKβ: IKKβ-htNSCs; IκBα: IκBα-htNSCs; GFP: GFP-htNSCs * P < 0.05, ** P < 0.01, *** P < 0.001, n = 4 (e,f,h,k), n = 5 (j) and n = 6 (m) per group. Error bars reflect means ± s.e.m. Statistics in (f,k): data points in red or blue lines were compared to the corresponding points in green line. Scale bar = 50 μm (d,g,i,l).

Figure 6

In vitro effect of IKKβ/NF-κB activation on neuronal differentiation of htNSCs. (a–e) Dissociated IKKβ-htNSCs, IκBα-htNSCs and GFP-htNSCs from Passage 6 were induced to differentiate. (a,b) Immunostaining of neuronal marker Tuj1. (b) The percentage of Tuj1-positive (Tuj1⁺) cells. IKKβ: IKKβ-htNSCs; IκBα: IκBα-htNSCs; GFP: GFP-htNSCs. (c,d) mRNA levels of genes encoding Notch isoforms (c) and Notch ligands (d). (e) Analysis of Notch signaling via Western blot measurement of cleaved Notch1 protein. (f–i) NF-κB controls genes that encode Notch signaling proteins. (f) NF-κB DNA-binding motif in the promoter regions of murine DLL4, Notch1 and Notch4 genes. (g–i) Gene promoter activities of wildtype (WT) vs. mutant (Mut) murine DLL4 (g), Notch1 (h) and Notch4 (i) in HEK293 cells in presence or absence of IKKβ/NF-κB activation, induced by transfection of pcDNA expressing constitutively-active IKKβ or a control (Con). (j,k) IKKβ-htNSCs were co-infected with lentiviral shRNAs against Notch1–4, as evaluated in Suppl. Fig. 8. Cells were induced to differentiate and analyzed for neuronal marker Tuj1 via immunostaining (j) and quantitatively analyzed for Tuj1-positive (Tuj1⁺) cells (k). * P < 0.05, ** P < 0.01, *** P < 0.001, n = 4 (b,c,d,g,h,i) and n = 6 (k) per group; error bars reflect means ± s.e.m. Scale bar = 50 μm (a,j).

Figure 7

Effect of NF-κB inhibition on differentiation of obese mice-derived htNSCs. (a,b) Dissociated IκBα-htNSCs^HFD, GFP-htNSCs^HFD and GFP-htNSCs^chow (established in Supplementary Fig. S7) at Passage 6 were induced to differentiate. Cells were then immunostained for neuronal marker Tuj1. (b) Percentage of Tuj1-positive (Tuj1⁺) cells. (c) Western blotting of cleaved Notch1. β-Actin (β-Act) was used to provide an internal control. (d,e) GFP-htNSCs^HFD expressing lentiviral Notch1–4 shRNAs vs. control shRNA were induced to differentiate for 7 days. Cells were then fixed and analyzed for Tuj1 Immunostaining (d). (e) Percentage of Tuj1-positive (Tuj1⁺) cells. ** P < 0.01, *** P < 0.001, n = 4 per group (b,e); error bars reflect means ± s.e.m. Scale bar = 50 μm (a,c,d).

Figure 8

Mouse model of htNSCs-specific IKKβ activation and metabolic phenotypes. C57BL/6 mice (3-month-old, chow-fed males) bilaterally received intra-MBH injections of P_Sox2-^CA IKKβ vs. P_Sox2-control lentiviruses. All mice were maintained under normal chow feeding throughout experiments. (a,b) Schematic of lentiviral vector that expressed ^CAIKKβ under the control of Sox2 promoter (P_Sox2-^CA IKKβ). The same vector with the removal of ^CAIKKβ was used as the matched control (P_Sox2-control). (b) Hypothalamic sections were prepared from mice at 2 weeks post injection and co-immunostained for Sox2 (green) and IκBα (red). (c–e) Mice with P_Sox2-^CAIKKβ vs. P_Sox2-control received a single-day icv injection of Brdu. Brains were fixed at Day 1, 7 or 14 post Brdu injection. Brain sections across the ARC were processed with Brdu staining or co-immunostaining with NeuN or S100B, and analyzed for total Brdu-labeled cells (c) and Brdu-labeled cells positive for NeuN (d) or S100B (e). (f–h) Brain sections across the ARC were prepared from mice at ~3 months (3 M) post lentiviral injection, subjected to Sox2 and POMC immunostaining, and counted for Sox2-positive cells (f), POMC neurons (g) and NPY neurons (h) in serial ARC sections. (i–l) Data show glucose tolerance (i) and fasting insulin levels (j) of mice at 3 M post lentiviral injection, and food intake (k) and body weight (l) of mice at 10 months (10 M) post lentiviral injections. Baseline body weight levels of mice prior to lentiviral injections were also included in (l). GTT: glucose tolerance test. * P < 0.05, ** P < 0.01, *** P < 0.001, n = 6 mice (c), n = 5 mice (c,e) and n = 4 mice (f,g,h) per group; n = 6 mice per group (d,i,j,k) and n = 10 mice per group (l). Error bars reflect means ± s.e.m. Scale bar = 50 μm (b).