Characterization of the Subventricular-Thalamo-Cortical Circuit in the NP-C Mouse Brain, and New Insights Regarding Treatment

Abstract

Gliosis in Niemann-Pick type C (NP-C) disease is characterized by marked changes in microglia and astrocytes. However, the gliosis onset and progression in NP-C has not been systematically studied, nor has the mechanism underlying this finding. Here, we found early gliosis in the subventricular zone (SVZ) of NP-C mice. Neural progenitor damage by Npc1 mutation suppressed vascular endothelial growth factor (VEGF) expression and further induced microglia activation followed by astrogliosis. Interestingly, excessive astrogliosis in the SVZ induced neural progenitor retention and/or migration into thalamus via astrocyte-derived VEGF, resulting in acceleration of thalamic and cortical gliosis through thalamo-cortical pathways. Transplantation of VEGF-overexpressing neural stem cells into the SVZ improved whole-brain pathology of NP-C mice. Overall, our data provide a new pathological perspective on NP-C neural pathology, revealing abnormalities in the subventricular-thalamo-cortical circuit of NP-C mouse brain and highlighting the importance of the SVZ microenvironment as a therapeutic target for NP-C disease.

Park et al. show early gliosis in the subventricular zone (SVZ) of Niemann-Pick type C mice. SVZ gliosis leads to abnormal migration of neural progenitors to the thalamus, and it accelerates thalamic gliosis and further cortical gliosis through thalamo-cortical pathways.

Introduction

Niemann-Pick type C (NP-C) disease is a rare and fatal neurovisceral storage disorder caused by mutations in the NPC1 gene.¹ In the NP-C brain, which is the major site of NP-C disease pathology, there is significant accumulation of gangliosides and sphingolipids, as well as axonal abnormalities, neurofibrillary tangles, and progressive neurodegeneration.² Previous studies have suggested that excessively reactive microglia and astrocytes may be causally related to the neuronal degeneration in NP-C disease. Also, many studies have shown pathological changes in microglia and astrocytes in the brains of NP-C patients and mice.3, 4, 5 However, despite comprehensive information regarding the genetic basis of NP-C disease, the underlying mechanism(s) leading to excessive gliosis in NP-C brain is not well understood.

In this study we focused on the subventricular zone (SVZ) of NP-C mice, widely known as a neurogenic niche where neural progenitors reside in the adult brain.⁶ The SVZ is responsible for adult neurogenesis by continuously generating new neurons from neural progenitors, and is important for maintaining the overall brain environment.6, 7 Prior work has reported that neural progenitors in the SVZ can leave their niche and migrate to various brain regions along stereotypical or non-stereotypical routes,8, 9, 10 supporting the significant role of the SVZ to regulate and maintain a healthy brain environment.

Here, we provide data demonstrating that suppression of vascular endothelial growth factor (VEGF) expression by Npc1 gene mutation reduced neural progenitor survival in the SVZ and contributed to early microglia activation through abnormal “off/on” signaling, mediating to astrogliosis. Interestingly, a subsequent increase of VEGF derived from reactive astrocytes, which are a source of chemoattractants in the brain,11, 12 affected abnormal migration of SVZ neural progenitors to the thalamus, resulting in accelerated thalamic gliosis of NP-C mice. In addition, excessive gliosis in the thalamus of NP-C mice mediated cortical gliosis through the interconnection of the thalamo-cortical pathway. Overall, these results indicate that the subventricular-thalamo-cortical circuit contributes to gliosis of the NP-C mouse brain. Treatment by injection of VEGF-overexpressing neural stem cells (NSCs) into the SVZ restored excessive gliosis throughout the brain of NP-C mice and improved NP-C mouse brain pathology, including changes in sphingolipids, cholesterol, and behavior. Therefore, our findings suggest that gliosis in NP-C mouse brain was related with an impairment of the SVZ microenvironment, and SVZ improvement by VEGF could be a potential therapeutic approach for NP-C disease.

Results

Early Gliosis in the SVZ of NP-C Mice

To understand the time course of gliosis changes in NP-C mice, we first examined change of astrocytes and microglia in the major brain regions, which have been previously shown to exhibit excessive gliosis in NP-C brain3, 4, 5 in an age-dependent manner. Glial fibrillary acidic protein (GFAP)-positive astrocytes were initially observed in thalamus of 2-week-old NP-C mice, while in cortex, amygdala, and brain stem they appeared at 4 weeks (Figure 1A). In contrast with astrocyte activation, the Iba-1 microglia were significantly increased in all brain regions of NP-C mice at 1 week of age. These results indicated that microglia activation began earlier than astrogliosis. Microglia activation was most severe in thalamus at 6-week-old NP-C mice, similar to astrogliosis (Figure 1B).

Figure 1.

Thalamus Is the Vulnerable Site of the Regions on Gliosis in NP-C Mouse Brain

(A and B) Representative images and quantification of astrocytes (GFAP) (A) and microglia (Iba-1) (B) in various brain regions of WT and NP-C mice with age (n = 6 mice per group). Scale bars, 20 μm. (C) Morphology of microglia in the cortex and thalamus of each group. Up, IMARIS-based three-dimensional reconstruction of a representative microglia. Scale bars, 10 μm. Down, IMARIS-based automated quantification of microglial morphology (n = 6 mice per group). (A and B) One-way ANOVA, Tukey’s post hoc test for comparison between the different regions at each age or the same region among the ages. (C) One-way ANOVA, Tukey’s post hoc test for comparison between the same region along the ages and Student’s t test for comparison between cortex and thalamus at each age. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM.

To more specifically assess microglia activation, we analyzed microglia morphology in cortex and thalamus, which exhibited increased microglia activation compared to other brain regions. Microglia in thalamus of 4-week-old NP-C mice showed larger cell bodies, longer processes, increased numbers of branching, and terminal points compared with cortex (Figure 1C), indicating that the thalamus is the most vulnerable site of the regions we have specifically looked at on microglia activation of the NP-C mouse brain.

We next assessed changes in gliosis within SVZ adjacent to the thalamus. Since GFAP is used as a marker of neural progenitors as well as activated astrocytes in SVZ,7, 13 we confirmed astrocyte activation by double staining with GFAP and SOX2. Both SOX2⁻GFAP⁺ astrocytes and microglia activation were already evident in SVZ of 1-week-old NP-C mice (Figures 2A and 2B). Morphological microglia activation also showed a marked early increase in SVZ (Figure 2C). Overall, these results demonstrated that the gliosis in NP-C mouse brain appeared earliest in SVZ compared to other brain regions.

Figure 2.

Early Appearance of Gliosis in the SVZ of NP-C Mouse Brains

(A and B) Representative images and quantification of SOX2⁻GFAP⁺ astrocytes (A) and Iba-1 (B) in the SVZ of each group (n = 6 mice per group). Scale bars, 20 μm. STR, striatum; LV, lateral ventricle. (C) IMARIS-based quantification of microglial morphology in the SVZ of each group. Scale bars, 10 μm (n = 6 per group). (A–C) One-way ANOVA, Tukey’s post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM.

Effect of the Neural Progenitor-Microglia-Astrocyte Triad on SVZ Gliosis in NP-C Mice

The SVZ is a specialized brain region in which neural progenitor subpopulations reside, such as type B, C, and A cells, and maintain the overall brain environment.6, 7, 13 We observed that the number of these neural progenitor subpopulations were reduced in the SVZ of NP-C mice compared to wild-type (WT) mice (Figure 3A). To more specifically investigate which cells in the NP-C SVZ were involved in gliosis and reduction of neural progenitor populations, Nestin-cre,¹⁴ CX₃CR1-creER,¹⁵ or GFAP-creER¹⁶ mice were crossed with NPC1^flox/flox mice¹⁷ to drive Npc1 gene deletion in neural progenitors, microglia, and astrocytes, respectively. To induce Cre activity in GFAP-creER; NPC1^flox/flox and CX₃CR1-creER; NPC1^flox/flox mice, tamoxifen (100 mg/kg) was intraperitoneally administered at postnatal day 30 for 5 days. Npc1 mRNA levels were markedly reduced in neural progenitors, microglia, and astrocytes isolated from SVZ of each mouse (Figure 3B). Next, we assessed microglia and astrocyte activation in the SVZ of these mice. The microglia were activated in SVZ of Nestin-cre; NPC1^flox/flox and CX₃CR1-creER; NPC1^flox/flox mice, but not in GFAP-creER; NPC1^flox/flox mice (Figures 3C and 3D). However, SOX2⁻GFAP⁺ astrocytes were increased in the SVZ of all mice compared with control mice (Figure 3E). These results showed that deficiency of Npc1 in neural progenitors could activate both microglia and astrocytes. Npc1 deletion in microglia also mediated microglia and astrocyte activation, while deletion in astrocytes did not affect microglia activation.

Figure 3.

Dysfunction of the Npc1 Gene in Neural Progenitors Induces Microglia Activation following Astrogliosis in the SVZ

(A) Left, representative immunofluorescence images of neural progenitor populations (type B cells, SOX2⁺ [green] and GFAP⁺ [red]; C cells, Mash1⁺ [green]; and A cells, DCX⁺ [green]) in the SVZ of WT and NP-C mice. Scale bars, 50 μm. Right, quantification of B, C, and A cells in the SVZ (n = 6 mice per group). (B) Npc1 mRNA in neural progenitors, microglia, and astrocytes isolated from the SVZ of control (+/+), Nestin-cre; NPC1^flox/flox, GFAP-creER; NPC1^flox/flox, and CX₃CR1-creER; NPC1^flox/flox mice (n = 6 mice per group). (C and D) Representative images and quantification of Iba-1 in the SVZ (C) and IMARIS-based automated quantification of microglial morphology (D) in the SVZ (n = 6 mice per group). STR, striatum; LV, lateral ventricle. (E) Representative images and quantification of SOX2⁻GFAP⁺ astrocytes in the SVZ (n = 6 mice per group). (F and G) Representative images and quantification of Iba-1 (F) and SOX2⁻GFAP⁺ astrocytes (G) in the SVZ of CX₃CR1-creER; NPC1^flox/flox and GFAP-creER; NPC1^flox/flox mice with age (n = 4 mice per group). (H) Quantification of B, C, and A cells in the SVZ (n = 6 mice per group). (A–E and H) The data analysis was done in 6-week-old WT and NP-C mice or 8-week-old control (+/+), Nestin-cre; NPC1^flox/flox, GFAP-creER; NPC1^flox/flox, and CX₃CR1-creER; NPC1^flox/flox mice. (A and B) Student’s t test. (C–H) One-way ANOVA, Tukey’s post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM. All expression levels are relative to Gapdh mRNA. See also Figures S1–S5.

Previous studies reported that neuronal damage involved in microglial activation by releasing so-called “off” and “on” signals.¹⁸ “Off” signals (BDNF, CD47, CD200, and CX₃CL1) are constitutively found in the healthy, normal functioning brain microenvironment and are responsible for keeping microglia in their resting state and antagonizing pro-inflammatory activity. In contrast, “on” signals (CXCL10, CCL21, and MMP3) are present under conditions of neuronal damage and can initiate detrimental microglia function. To determine the correlation between neuronal damage with microglia activation in the SVZ of NP-C and Nestin-cre; NPC1^flox/flox mice, we first confirmed the expression level of the neuronal-specific gene. The neuronal-specific markers Tuj1, GAP43, MBP, and MAP2 were decreased in neural progenitors from SVZ of NP-C and Nestin-cre; NPC1^flox/flox mice (Figure S1A). Moreover, neuronal damage in these mice decreased “off” signals and increased “on” signals compared with WT or control mice, mediating to upregulation of pro-inflammatory cytokines, especially interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α), in microglia from SVZ of these mice (Figures S1B and S1C). These findings indicated that Npc1-deficient neural progenitors mediated microglia activation through dysregulation of “off/on” signals in the SVZ.

To validate microglia-mediated astrocyte activation, we examined the change in the activation of microglia and astrocytes in the SVZ of CX₃CR1-creER; NPC1^flox/flox and GFAP-creER; NPC1^flox/flox mice in an age-dependent manner. CX₃CR1-creER; NPC1^flox/flox mice demonstrated an increase of the activated microglia at all ages, but astrocyte activation appeared after microglia activation. In addition, GFAP-creER; NPC1^flox/flox mice demonstrated activation of astrocytes, while microglia were not activated (Figures 3F and 3G). Thus, these data substantiate the fact that microglia activation came first and then astrogliosis in the SVZ of NP-C mice occurred. Neural progenitor populations were decreased in Nestin-cre; NPC1^flox/flox mice; however, CX₃CR1-creER; NPC1^flox/flox mice showed no significant difference (Figure 3H), suggesting that microglia damage by Npc1 deletion did not influence neural progenitor survival in the SVZ. Collectively, these data determined that dysfunction of Npc1 in neural progenitors induced microglia activation and activated microglia-mediated astrogliosis in the SVZ.

VEGF-Deficient Neural Progenitors in SVZ Gliosis of NP-C Mice

VEGF, which has been described as an endothelial cell-specific mitogen, also affects neuronal survival and neuroinflammation in the brain.19, 20 Moreover, we previously observed suppression of VEGF activity in Purkinje neurons of NP-C mice.²¹ To confirm whether VEGF in neural progenitors mediates SVZ gliosis in NP-C mice, we assessed the expression level of VEGF in neural progenitors within the SVZ of NP-C in an age-dependent manner. The result showed that the reduction of VEGF expression in the SVZ neural progenitors was initiated in 1-week-old NP-C mice (Figure S2A). Additionally, we observed decrease of VEGF expression in the SVZ neural progenitors of Nestin-cre; NPC1^flox/flox mice (Figure S2B). The immunofluorescence staining of SOX2 and VEGF in the SVZ also showed reduction of VEGF in neural progenitors in NP-C and Nestin-cre; NPC1^flox/flox mice (Figure S2C). The Nestin-cre; VEGF^flox/flox mice,²² which are conditional knockout mice lacking VEGF in neural progenitors, showed upregulation of microglia and astrocyte activation and decrease of neural progenitor populations in the SVZ, similar to NP-C or Nestin-cre; NPC1^flox/flox mice (Figures S2D–S2H). These observations were related to the neuronal damage and change of “off/on” signals in neural progenitors mediated by VEGF deficiency and resulted in microglia activation by increasing pro-inflammatory cytokines such as IL-1β (Figures S2I–S2K). Overall, these results suggested that VEGF in neural progenitors mediated SVZ gliosis in NP-C mice by regulating “off/on” signals.

To further confirm VEGF-mediated SVZ gliosis through “off/on” signals in vitro, we isolated primary neural progenitors and microglia from the SVZ of WT, NP-C, control, Nestin-cre; NPC1^flox/flox, or Nestin-cre; VEGF^flox/flox mice. The conditioned media (CM) derived from SVZ neural progenitors of each mouse type was harvested and transferred to microglia cultured from the SVZ of WT mice (Figure S3A). The CM derived from SVZ neural progenitors of NP-C, Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mice contained lower levels of “off” signal cytokines and high levels of “on” signal cytokines than CM derived from SVZ neural progenitors of WT and control mice (Figure S3B). These changes of “off/on” signal cytokines in CM induced microglia activation by significantly increasing IL-1β and TNF-α levels (Figures S3C–S3E). Moreover, the CM derived from SVZ microglia of these mice induced elevation of specific genes (GFAP, Vimentin, and Lipocalin 2 [LCN2]) related with astrocyte activation, resulting in morphological astrocyte activation (Figures S4A–S4C).

Next, to more exactly define the neural progenitor-microglia-astrocyte triad on gliosis of NP-C mice, we performed sequential CM treatment. The CM derived from SVZ neural progenitors of WT, NP-C, +/+, Nestin-cre; NPC1^flox/flox, or Nestin-cre; VEGF^flox/flox mice were applied to WT microglia, and then the CM from these microglia were transferred to WT astrocytes. These data indicated that VEGF-deficient neural progenitors (due to Npc1 dysfunction) induced astrogliosis through mediation of microglia activation (Figures S5A–S5C). To confirm whether neural progenitors directly induce astrogliosis and further it mediate microglia activation, the CM derived from SVZ neural progenitors of these mice were applied to WT astrocytes, and then the CM from these astrocytes were transferred to WT microglia. We observed that CM derived from the SVZ neural progenitors of these mice did not directly induce astrocyte activation, nor did they induce microglia activation (Figures S5D–S5F). Taken together, SVZ astrogliosis in NP-C mouse brain occurred by microglia activation and was regulated through the “off/on” signals secreted from VEGF-deficient SVZ neural progenitors.

Abnormal Retention and/or Migration of SVZ Neural Progenitors in NP-C Mouse Brain

Neural progenitors in the SVZ migrate tangentially through the rostral migratory stream (RMS) to the olfactory bulb (OB), where they differentiate into neurons. Various signaling pathways regulate this neural progenitor migration pattern,7, 23 and previous studies have reported that VEGF is one of the regulators implicated in this process.24, 25 We first confirmed that SVZ neural progenitors in NP-C mice showed reduction of VEGF. Based on this concept and result, we then investigated migration of neural progenitors in WT and NP-C mice. To visualize endogenous neural progenitor migration along the RMS, we first confirmed successful labeling and no differences of incorporation efficiency between WT and NP-C SVZ neural progenitors in vitro using micron-sized iron oxide particles (MPIOs) (Figure 4A), and then mice were injected with MPIOs in the lateral ventricle (LV) for labeling of endogenous neural progenitors²⁶ (Figure 4B). Seven days after intraventricular injection of MPIOs, MPIO⁺ neural progenitors were detected in the OB of WT mice; in comparison, NP-C mice exhibited reduction of MPIO⁺ neural progenitor migration into the OB (Figure 4C).

Figure 4.

NP-C Mice Exhibit Abnormal Migration of Neural Progenitors to the Thalamus

(A) Quantification of MPIO-labeled SVZ neural progenitors in vitro (n = 5 independent experiments). (B) Experimental design to investigate neural progenitor migration. (C) Left, representative immunofluorescence images of MPIO⁺ neural progenitor (green) migration. Scale bars, 20 μm. Right, quantification of MPIO⁺ neural progenitors in WT and NP-C mouse brain regions at 7 days (upper) or 21 days (lower) (n = 6 mice per group). (D) Representative immunofluorescence images of MPIO⁺ neural progenitors merged with DCX (migrating neural progenitors, red) in the SVZ, RMS, and OB of WT mice and the thalamus of NP-C mice. Scale bars, 20 μm. (E) Flow cytometry histogram of MPIO⁺ cells incorporated with DCX or NeuN in the thalamus of NP-C mice. (F) Number of NSs derived from the SVZ and thalamus (n = 6 mice per group). (G) Quantification of MPIO⁺ neural progenitors in each part of control (+/+), Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mouse brains (n = 6 mice per group). The data analysis was done in 6-week-old WT and NP-C mice or to 8-week-old control (+/+), Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mice except (C). (A and G) One-way ANOVA, Tukey’s post hoc test. (C and F) Student’s t test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM. See also Figure S6.

Interestingly, these NP-C cells migrated along the lower part of the LV (LLV) and were finally observed in the thalamus 21 days after MPIO injection (Figure 4C; Figure S6A). To confirm whether the MPIOs were incorporated in migrating neural progenitors, we performed double staining with doublecortin (DCX) (migrating neural progenitors) in the SVZ, RMS, and OB. MPIO⁺ neural progenitors were observed in DCX⁺ cells in the three parts of the SVZ-RMS-OB migratory pathway in WT mice and also in the thalamus of NP-C mice (Figure 4D). To more clearly determine the fate of neural progenitors migrated into the thalamus of NP-C mice, MPIO⁺ cells from thalamus were stained with DCX or NeuN (neuron marker) for flow cytometry analysis. The results showed that most of MPIO⁺ cells were positive for DCX but not NeuN staining in the thalamus (Figure 4E), indicating that neural progenitors that migrated into the thalamus of NP-C mice remained in a neural progenitor-like state.

We further confirmed migration of neural progenitors to the thalamus in NP-C mice by performing neurosphere (NS) assays on tissue derived from the SVZ and thalamus. Similar to previous results, NSs surprisingly formed within the thalamus of NP-C mice compared with WT mice (Figure 4F). Nestin-cre; NPC1^flox/flox also showed abnormal migration of neural progenitors into the thalamus. Although this unusual migration pattern was not exhibited in Nestin-cre; VEGF^flox/flox mice, these mice showed retention of MPIO⁺ neural progenitors in LV and LLV compared with control mice, resulting in reduction of MPIO⁺ neural progenitors in the OB (Figure 4G; Figure S6B). Overall, these results demonstrated that deficiency of VEGF in neural progenitors contributed to their decreased migration to the OB in NP-C mice. Moreover, for the first time, they revealed the migration of neural progenitors to the thalamus in NP-C mice brain, strongly suggesting that this abnormal migration could affect thalamic gliosis of NP-C mice.

Astrocyte-Derived VEGF as a Chemoattractant of Neural Progenitors in the SVZ

CNS injury, hypoxia, microglial activation, and the subsequent release of inflammatory cytokines invoke complex responses known collectively as astrogliosis.12, 27 The activated astrocytes are a source of chemoattractants (such as VEGF, stromal cell-derived factor 1 [SDF1], monocyte chemoattractant protein 1 [MCP1], fibroblast growth factor 2 [FGF2], and insulin-like growth factor 1 [IGF1]) that act on neural progenitors. These factors, released from activated astrocytes, direct neural progenitor chain migration along non-stereotypical routes.11, 12, 28 As described above, we observed elevation of pro-inflammatory cytokines (IL-1β and TNF-α), abnormal SVZ neural progenitors and neural progenitor migration, decreased VEGF due to Npc1 dysfunction, and ultimately reactivated astrocytes. We therefore hypothesized that chemoattractants secreted from reactive astrocytes by pro-inflammatory cytokines induced retention of neural progenitors within the SVZ microenvironment and consequently affected migration of neural progenitors into the thalamus, which is contiguous with SVZ. To therefore determine the effects of reactive astrocytes on neural progenitor attraction in vitro, we performed migration assays. Notably, neural progenitors exhibited robust mobility toward astrocytes exposed to SVZ microglia CM derived from NP-C, Nestin-cre; NPC1^flox/flox or Nestin-cre; VEGF^flox/flox mice (Figure 5A).

Figure 5.

Astrocyte-Derived VEGF Contributes to Abnormal Retention and/or Migration of Neural Progenitors

(A) Illustration of the transwell migration assay (left) and comparison of the migration ability (right) of WT SVZ neural progenitors by astrocytes exposed to CM derived from SVZ microglia of WT, NP-C, control (+/+), Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mouse brains (n = 6 independent experiment). (B) mRNA levels of chemoattractants in astrocytes (n = 6 independent experiments). (C) Protein level of VEGF secreted from astrocytes (n = 6 independent experiments). (D) Illustration of IL-1β (10 ng/mL) or IL-1β antibody (10 ng/mL) treatment in astrocytes (left) and mRNA levels of chemoattractants (right) after 24 h (n = 6 independent experiments). (E) Protein level of VEGF secreted from astrocytes (n = 6 independent experiment). (F) Illustration of the transwell migration assay (left) and comparison of the migration ability (right) of WT SVZ neural progenitors by control astrocytes treated with IL-1β or IL-1β antibody (n = 6 independent experiments). (G) Illustration of the transwell migration assay (upper) and comparison of the migration ability (lower) of WT SVZ neural progenitors by control or VEGF knockdown astrocytes with or without IL-1β treatment (n = 6 independent experiments). (H) Upper, representative immunofluorescence images of GFAP (green) and VEGF (red) in the SVZ and LLV of each group. Scale bars, 50 μm. Lower, quantification of total VEGF (left) and astrocyte-derived VEGF (GFAP⁺VEGF⁺, right) (n = 6 mice per group). (I) Representative flow cytometry plot (left) and percentage (right) of astrocyte-derived VEGF (GLAST-1⁺ CD11b⁻ VEGF⁺) in the SVZ (n = 6 mice per group). (J) mRNA levels of VEGF in astrocytes sorted from the SVZ (n = 6 mice per group). All data analysis was done in 6-week-old WT and NP-C mice or 8-week-old control (+/+), Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mice. (A–C, I, and J) Student’s t test between WT and NP-C mice. One-way ANOVA, Tukey’s post hoc test among control (+/+), Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mice. (D–F) One-way ANOVA, Tukey’s post hoc test. (G) Student’s t test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM. All expression levels are relative to Gapdh mRNA.

Next, we determined the levels of chemotactic factors in astrocytes of each group. An incremental expression and secretion of chemotactic factors, particularly VEGF, was detected in the astrocytes exposed to SVZ microglia CM (Figures 5B and 5C). These results were supported by the fact that VEGF has recently been shown to express strongly in astrocytes in response to IL-1β secreted from microglia.²⁹ To validate the exact relationship between IL-1β and VEGF elevation in astrocytes, astrocytes were respectively treated with IL-1β and IL-1β antibody. VEGF secretion increased in astrocytes with IL-1β treatment, but not with IL-1β antibody treatment (Figures 5D and 5E). In addition, astrocyte-derived VEGF increased following IL-1β treatment, which mediated the migration of neural progenitors. IL-1β antibody, however, had no effect on neural progenitor migration (Figure 5F). VEGF knockdown in astrocytes also did not induce migration despite IL-1β treatment (Figure 5G). The elevation of astrocyte-derived VEGF was further confirmed in the SVZ or LLV of NP-C, Nestin-cre; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mice through immunofluorescence staining, flow cytometry, and astrocyte sorting (Figures 5H–5J). Together, these data suggested that VEGF, which emanated from early astrogliosis in the SVZ of NP-C mice, contributed chemoattractive retention of neural progenitors within the SVZ microenvironment and resulted in abnormal migration into the thalamus. Further, it might impact acceleration of thalamic gliosis in NP-C mice.

Characterization of the Subventricular-Thalamo-Cortical Circuit in NP-C Mouse Brain

Based on the above findings, we further investigated whether SVZ impairment, such as excessive SVZ gliosis in NP-C mice, affects thalamic gliosis through the abnormal migration of neural progenitors. To locally deplete of Npc1 or Vegf only in the SVZ, we daily injected Split-cre, which specifically drives excision in GFAP⁺CD133⁺ neural progenitors,³⁰ into the LV of NPC1^flox/flox and VEGF ^flox/flox mice (Figure 6A). After 4 weeks, we confirmed that Npc1 or Vegf mRNA levels were markedly reduced in neural progenitors isolated from the SVZ of Split-cre-injected NPC1^flox/flox and VEGF ^flox/flox mice (Figure 6B). NPC1^flox/flox and VEGF ^flox/flox mice injected with Split-cre showed reduction of neural progenitor populations and significant gliosis in SVZ (Figures 6C and 6D). Astrogliosis in the SVZ of these mice resulted in an increase of astrocyte-derived VEGF and mediated retention/abnormal migration of Npc1 or VEGF-reduced neural progenitors to the thalamus (Figures 6E–6G). This abnormal migration of defected neural progenitors finally induced gliosis in the thalamus (Figure 6H). We also reconfirmed abnormal migration of these neural progenitors visually in Split-cre-injected NPC1^flox/flox; td-tomato^flox/flox mice³¹ (Figure 6I). Thus, these results indicated that thalamic gliosis could be induced by abnormal migration of Npc1- or VEGF-depleted neural progenitors from the SVZ.

Figure 6.

SVZ Impairment Affects Gliosis in the Thalamus and Cortex

(A) Experimental design to investigate the effect of the NP-C SVZ defect on brain gliosis. GFAP/CD133 Split-cre (1/1.5 ratio) was injected daily into the LV of NPC1^flox/flox and VEGF^flox/flox mice (3 weeks old) for 4 weeks. MPIOs were injected 1 week after Split-cre injection. (B) Npc1 and Vegf mRNA in neural progenitors isolated from the SVZ of GFAP/CD133 Split-cre LV-injected NPC^flox/flox and VEGF^flox/flox mice (n = 4 mice per group). (C–E) Quantification of B, C, and A cells in the SVZ (C), Iba-1 and GFAP (D) in the SVZ, and astrocyte-derived VEGF (GFAP⁺VEGF⁺) (E) in the SVZ and LLV (n = 6 mice per group). (F) Quantification of MPIO⁺ neural progenitors in each part of the Split-cre-injected mouse brain (n = 6 mice per group). (G) Number of NSs derived from the SVZ and thalamus (n = 6 mice per group). (H) Quantification of Iba-1 and GFAP in the thalamus (n = 6 mice per group). (I) Left, representative immunofluorescence images of td-tomato⁺ neural progenitor (red) migration of Split-cre-injected NPC1^flox/flox or NPC1^flox/flox; td-tomato^flox/flox mice. Scale bars, 50 μm. Right, quantification of td-tomato⁺ neural progenitors in each brain region (n = 6 mice per group). (J) Quantification of Iba-1 and GFAP in the cortex (n = 6 mice per group). (K) Number of NSs derived from the cortex (n = 6 mice per group). (L and M) Unesterified cholesterol (L), sphingosine, and sphingomyelin (M) levels in cortex (n = 6 mice per group). (N) Quantification of sensory function by hot-plate and tail-flick tests. Latency indicates the amount of time until the mouse showed signs of jumping or paw licking for hot-plate test, or tail flick (n = 10–13 mice per group). (O) Left, time spent in the wall side, time spent in center regions and percent of center region (upper), and representative traces of mouse movement (lower) during open-field test (n = 10–13 mice per group). Right, rotarod scores and beam test (12- or 6-mm square beam) (n = 10–13 mice per group). All data analysis was done in 7- to 8-week-old mice. (B and I) Student’s t test. (C–H and J–O) One-way ANOVA, Tukey’s post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM. All expression levels are relative to Gapdh mRNA. See also Figure S7.

Interestingly, this SVZ impairment also contributed to gliosis in the cortex (Figure 6J). However, this was not related to migration of defected neural progenitors from the SVZ to the cortex (Figure 6K), indicating that thalamic gliosis itself affected cortical gliosis. Moreover, Split-cre injected NPC1^flox/flox and VEGF ^flox/flox mice showed cholesterol and sphingolipid accumulation in the cortex, which contributes to the major brain pathology of NP-C mice, resulting in alteration of behaviors related with sensory and motor function (Figures 6L–6O). Therefore, these findings highlight the importance of the SVZ on regulation of the whole-brain environment in NP-C mice.

The thalamus relays sensori-motor information to the cortex and is an integral part of cortical executive functions.³² Previous studies demonstrated that thalamus inflammation affects the cortex along the thalamo-cortical pathway.5, 32, 33 In our results above, we observed that the SVZ neuronal damage induced gliosis and mediated to abnormal migration of Npc1- or VEGF-depleted neural progenitors to the thalamus, resulting in thalamic gliosis and further cortical gliosis. To examine whether cortical gliosis was accompanied with neuron loss through the thalamo-cortical pathway, we examined neuron loss in the thalamic nuclei (ventral posteromedial nucleus [VPM], ventral posterolateral nucleus [VPL], dorsal lateral geniculate nucleus [LGNd]) and cortical regions (primary somatosensory cortex barrel field [S1BF]) in SVZ-impaired mice by Spilt-cre LV injection. The data showed reduction of neurons in thalamic nuclei and cortical regions (Figures S7A and S7B). We confirmed these observations by unilateral injection of Npc1 short hairpin RNA (shRNA) into the thalamus of WT mice. In addition, we also transplanted NSCs cultured from the SVZ of NP-C mice into WT mice (Figure S7C). Npc1 expression was reduced in the thalamus by Npc1 shRNA injection (Figure S7D). Gliosis was observed in both the thalamus and cortex of Npc1 shRNA or NP-C NSC-injected mice (Figure S7E). Next, we analyzed neuron loss relative to the time course of thalamic injection. The results showed that neuron loss was already evident in the three thalamic nuclei at 2 weeks after injection of Npc1 shRNA or NP-C NSCs compared to control shRNA or WT NSC injection (Figure S7F). In contrast, neuron loss in the S1BF of the cortex was a later event compared with thalamic nuclei (Figure S7G). These results indicated that cortical gliosis, which was induced in the thalamic-defected mice by Npc1 shRNA or NP-C NSC injection, is induced by the interconnection of thalamo-cortical pathways. Taken together, our data showed that abnormal migration from the SVZ to the thalamus of VEGF-deficient neural progenitors due to Npc1 mutation mediated to thalamic gliosis and contributed to cortical gliosis through the thalamo-cortical pathway, suggesting an important effect of the subventricular-thalamo-cortical circuit on gliosis of NP-C mice brain.

Improvement of NP-C Mouse Brain Pathology by Replenishment of VEGF from NSCs in the SVZ

To investigate whether local VEGF administration restores NP-C mouse brain pathology, we injected WT NSCs, VEGF^tg NSCs, or VEGF-overexpressing vectors (VEGF OX vector) into the LV of NP-C mice. We first confirmed survival of transplanted VEGF^tg NSCs to determine the number of injections. MPIO-labeled VEGF^tg NSCs were injected into the LV of NP-C mice, and brain was collected after 3, 7, or 14 days. The results showed that most of MPIO-labeled VEGF^tg NSCs transplanted into the LV of NP-C mice were negative for DAPI staining at 3 days after injection. However, DAPI-positive cells (57%) were increased at 7 days after injection, and all of these cells (90%) were positive for DAPI staining at 14 days after injection. Moreover, we also found that the VEGF production of these cells gradually decreased with the day after injection (Figure S8). Therefore, these results indicated that transplanted VEGF^tg NSCs into the LV could maintain their survival and VEGF production maximum for 7 days in NP-C mouse brain. Based on this result, treatments were performed twice a week for 4 weeks to definite therapeutic effects on NP-C mice brain pathology (Figure 7A). Interestingly, reduction of neural progenitors in the SVZ and resolution of gliosis in the overall brain regions were improved only in VEGF^tg NSC-injected NP-C mice (Figures 7B–D). In addition, these mice showed lower levels of astrocyte-derived VEGF and correction of abnormal migration to the thalamus (Figures 7E–7G). To confirm that the injected VEGF^tg NSCs differentiate into neurons or remain in a less-differentiated progenitor-like state in the OB and thalamus, MPIO-labeled VEGF^tg NSCs were injected into the LV of NP-C mice, and then single-cells from OB and thalamus were stained with DCX or NeuN. The result showed that the most of the MPIO-positive cells were positive for DCX but not NeuN staining in both the OB and thalamus (Figure 7H). These results indicated that injected VEGF^tg NSCs were maintained in a less-differentiated progenitor-like state in migrated regions and suggested that VEGF secreted from the NSC state showed therapeutic effects. Next, we observed less neuronal loss in the thalamus and cortex of mice treated with LV injection of VEGF^tg NSCs compared to the PBS-treated group (Figure 7I). In addition, treatment with VEGF^tg NSC was associated with lower levels of cholesterol and sphingolipid in cortical neurons and astrocytes of NP-C mouse brain (Figures 7J–7L). These data, showing reduction of cholesterol and lipid accumulation in neurons, supported the fact that glia cells, especially astrocytes, are a likely source of cholesterol supply in neurons of the adult brain.34, 35 Finally, VEGF^tg NSC injection into NP-C mice mediated improvement of sensory and motor function and lifespan (Figures 7M–7O). Collectively, these data demonstrated that SVZ mitigation by VEGF^tg NSC LV injection could improve NP-C brain pathologies.

Figure 7.

LV Injection of VEGF^tg NSCs Improves NP-C Mouse Brain Pathology

(A) Experimental design to determine the therapeutic effect of local VEGF administration in LV of NP-C mice. (B and C) Quantification of B, C, and A cells (B) in the SVZ and Iba-1 and GFAP (C) in the SVZ (n = 6 mice per group). (D and E) Quantification of Iba-1 and GFAP (D) in the thalamus and cortex and astrocyte-derived VEGF (GFAP⁺VEGF⁺) (E) in SVZ and LLV (n = 6 mice per group). (F) Quantification of MPIO⁺ neural progenitors in each mouse brain region (n = 6 mice per group). (G) Number of NSs derived from the SVZ and thalamus (n = 6 mice per group). (H) Flow cytometry histogram of MPIO-labeled VEGF^tg NSCs incorporated with DCX or NeuN in the OB and thalamus. (I) The number of neurons in thalamic nuclei (VPM, VPL, LGNd) and cortical regions (S1BF) (n = 6 mice per group). (J) Unesterified cholesterol level by filipin staining in cortex (n = 6 mice per group). (K and L) Unesterified cholesterol (K), sphingosine, and sphingomyelin (L) levels in astrocytes and neurons isolated from the cortex (n = 6 mice per group). (M) Quantification of sensory function by hot-plate (upper) and tail-flick (lower) test (n = 10–12 mice per group). (N) Left, time spent in the wall side, time spent in center regions and percent of center region (upper), and representative traces of mouse movement (lower) during open-field test (n = 10–12 mice per group). Right, rotarod scores and beam test (12- or 6-mm square beam) (n = 10–12 mice per group). (O) Survival curve of each group (n = 12 mice per group). All data analysis was done on 9-week-old mice. (B–G and I–N) One-way ANOVA, Tukey’s post hoc test. (O) Log rank test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM. See also Figure S8.

2-hydroxypropyl-β-cyclodextrin (CD) has been demonstrated to dramatically ameliorate liver cholesterol storage and slightly delay neurological symptoms in Npc1 mutant mice.³⁶ However, it has clinical limitations, since it cannot permeate the blood-brain barrier.³⁷ To overcome this limitation, some studies have demonstrated the therapeutic effects using the intracerebral or intrathecal delivery of CDs.38, 39 Additionally, many researchers have tried combination therapies utilizing CD, miglustat, allopregnanolone, ibuprofen, and curcumin to increase therapeutic efficacy.40, 41 Ibuprofen, which is one of the non-steroidal anti-inflammatory drugs (NSAIDs) used in the treatment of CNS disease, reduced inflammation in the NP-C mouse brain, resulting in a slight increase in lifespan. These therapeutic effects resulted in a synergistic improvement on lifespan when combined with miglustat.41, 42 In reference to a concept in a previous study, we performed experiments to compare the beneficial effects of combination therapy focused on SVZ mitigation with CD, which has more therapeutic effects than miglustat. Mice were given a subcutaneous injection of CD (4,000 mg/kg) starting at 1 week old (once a week), intraventricular injection of VEGF^tg NSCs starting at 4 weeks old (twice a week), or ibuprofen (100 mg/kg/day) as a dry admixture to mouse chow starting at 6 weeks old due to the toxicity seen with earlier dosing (Figure 8A). The results demonstrated that compared to ibuprofen treatment, LV injection of VEGF^tg NSCs significantly reduced gliosis in the SVZ, thalamus, and cortex. Further, the effects of VEGF^tg NSCs combined with CD were greater than ibuprofen combined with CD. Monotherapy of CD also reduced gliosis in each brain region, but its effects were lower than those seen with combination therapy of VEGF^tg NSCs with CD (Figure 8B). The accumulation of cholesterol and sphingolipid in the NP-C mouse brain was unchanged by ibuprofen treatment. However, it did decrease with treatment of VEGF^tg NSCs, compared with CD monotherapy. Further, the combination of VEGF^tg NSCs and CD showed greater synergistic effect, compared to the combined treatment with ibuprofen and CD treatment (Figures 8C and 8D). These results demonstrated that combination therapy of VEGF^tg NSCs and CD was superior in improving brain pathology in NP-C mice. Moreover, the synergistic effects of VEGF^tg NSCs with CD resulted in greater enhancement of motor function and body weight than combined therapy of ibuprofen with CD (Figures 8E and 8F). In addition, these combination therapies were also clearly more beneficial to the lifespan (∼17 weeks) than with a combined treatment of ibuprofen with CD (∼15 weeks) or CD treatment alone (∼14 weeks) (Figure 7G). Collectively, these results indicated that NP-C brain pathology was most improved with combined VEGF^tg NSC LV injection and CD treatment than anti-inflammatory drugs such as ibuprofen with CD treatment. Therefore, the combined therapeutic strategy acting on improving the SVZ, together with existing NP-C treatment drugs, has substantial value for the improvement of NP-C disease. This highlights the importance of the SVZ microenvironment as a potential new therapeutic target for NP-C disease.

Figure 8.

Combination Therapy of VEGF^tg NSCs with CD Is More Beneficial for Improving NP-C Pathologies Compared to Ibuprofen Treatment Combined with CD

(A) Experimental design to determine the effect of combination therapy of VEGF^tg NSCs or ibuprofen with CD. (B) Quantification of Iba-1 and GFAP (n = 4 mice per group). (C) Unesterified cholesterol levels in the cortex, liver, lung, kidney, and spleen (n = 4 mice per group). (D) Sphingosine and sphingomyelin levels (n = 4 mice per group). (E–G) Rotarod scores (E), average body weight (F), and survival curve (G) of WT and NP-C mice in each treatment group (n = 10 mice per group). All data analysis was done with 10-week-old mice. (B–F) One-way ANOVA, Tukey’s post hoc test. (G) Log rank test. *p < 0.05, **p < 0.01, ***p < 0.001. All error bars indicate SEM.

Discussion

In this study, we revealed new insights regarding the subventricular-thalamo-cortical circuit and its influence on NP-C brain pathology in mice. The suppression of VEGF in neural progenitors by Npc1 gene mutation resulted in the reduction of neural progenitor survival in the SVZ. In addition, this induced lower levels of “off” signal and higher levels of “on” signal molecules, resulting in microglial activation and elevated pro-inflammatory cytokines, especially IL-1β. IL-1β secreted from reactive microglia subsequently mediated astrogliosis. The increase in astrocyte-derived VEGF contributed to the retention and/or migration of defective neural progenitors into the thalamus in NP-C mice. The abnormal migration of defective neural progenitors accelerated gliosis in the thalamus, further contributing to cortical gliosis by inducing cortical neuronal loss through the thalamo-cortical pathway. Based on these observations, the subventricular-thalamo-cortical circuit is proposed to underlie excessive gliosis in the NP-C mouse brain, particularly in the thalamus. Although we only focused on SVZ-mediated whole-brain gliosis in this study, a previous study showed reactive gliosis in the thalamus in the early stage of NP-C disease and subsequently in interconnected cortical laminae at later ages,⁵ supporting our results.

We have previously found that VEGF activity is reduced in NP-C cells including neurons and causes defective autophagy by sphingolipid change related with abnormalities in the VEGF/sphingosine kinase (SphK) pathway, resulting in progressive neuronal loss.²¹ In this study, we reconfirmed reduction of VEGF by Npc1 mutation in neural progenitors, resulting in a decrease of neural progenitor survival in the SVZ. As in our previous findings, this observation may be associated with VEGF-mediated abnormal autophagy. Moreover, Npc1-deficient mediated VEGF suppression may be related with hedgehog (Hh) signaling required for proper development, because VEGF expression has been known to depend on Hh signaling, and Hh receptors share significant homology with the NPC1 protein.43, 44 Thus, these previous studies suggested that Npc1 dysfunction is associated with impairment of Hh signaling. We are currently conducting further studies to determine the relationship between VEGF suppression in NP-C cells and Hh signaling.

Adult neurogenesis is responsible for the continuous generation and integration of new neurons in existing neural networks. These neurons are derived from multipotent neural progenitors residing in specialized neurogenic niches. The SVZ is the largest of these niches in the adult brain and is capable of generating new neurons that then migrate through the RMS to integrate into networks in the OB.6, 7 Interestingly, we found reduced migration of VEGF-suppressed neural progenitors to the OB and abnormal migration to the thalamus in NP-C mice brain. Previously, VEGF/VEGF receptor (VEGFR) signaling has been shown to mediate axonal guidance and formation of glia tube structures surrounding chains of neural progenitors for migration.24, 25 We confirmed that decrease of migration to the OB was related to abnormality of glia tube structure in the RMS of the NP-C mouse brain and was linked to impairment of olfactory sense (data not shown), suggesting a role of VEGF/VEGFR signaling as an important regulator of the neural progenitor-glia tube interaction during migration, consistent with previous studies. In addition, our results showed that VEGF in neural progenitors mediated its survival and gliosis, and astrocyte-derived VEGF could attract neural progenitors, indicating a bimodal role of VEGF with cell types in the NP-C mouse brain.

In this study, we focused on retention and/or migration of neural progenitors to the thalamus by astrocyte-derived VEGF in the SVZ and how this relates with excessive gliosis in thalamus and cortex. Recent findings, which characterized human ventriculo-olfactory neurogenic system (VONS) containing the SVZ, the RMS, the olfactory tract, and the OB, revealed that the human RMS is organized around a lateral ventricular extension reaching the OB, similar to that of rodents.¹⁰ Therefore, future studies focusing on NPC1 patient brain are needed to investigate the abnormal migration with regard to whole-brain gliosis, as observed in NP-C mice. Specific impairment of the SVZ microenvironment by LV injection of Split-cre in NPC1^flox/flox or VEGF^flox/flox mice exhibited abnormal brain pathology similar to NP-C mice. SVZ gliosis in these mice contributed to thalamic gliosis by abnormal migration of defected neural progenitors and mediated cortical gliosis through the thalamo-cortical pathway, suggesting the remarkable importance of the SVZ for improvement of the whole brain environment in NP-C mice.

Intraventricular transplantation of VEGF^tg NSCs improved NP-C mice brain pathology more than WT NSCs and VEGF OX vector, demonstrating that improving the SVZ microenvironment using cell therapy simultaneously with gene (overexpression or replacement) therapy is required to treat NP-C brain pathology. Combination therapy of VEGF^tg NSCs together with CD is more clearly beneficial in improving brain pathology, behavior, and further lifespan than other monotherapy or combination therapy, suggesting that this stem cell therapeutic strategy by VEGF focusing on SVZ mitigation may provide new opportunities in the treatment of NP-C disease.

Materials and Methods

Mice

A colony of BALB/c Npc1^nih mutant mice has been maintained, and the genotype of each mouse was determined by PCR as described.⁴⁵ Nestin-cre¹⁴ (The Jackson Laboratory), CX₃CR1-creER¹⁵ (The Jackson Laboratory), GFAP-creER¹⁶ (The Jackson Laboratory), NPC1^{flox/flox 17}, or VEGF^{flox/flox 22} mice were used to delete the Npc1 gene or VEGF in neural progenitors, astrocytes, and microglia, respectively. To induce Cre activity in GFAP-creER; NPC1^flox/flox and CX₃CR1-creER; NPC1^flox/flox, tamoxifen (100 mg/kg) was intraperitoneally administered at postnatal day 30 for 5 days. For visualization of neural progenitor cell migration, td-tomato^flox/flox mice³¹ (The Jackson Laboratory) were used. Transgenic mice overexpressing VEGF⁴⁶ in the brain under the control of neuron-specific promoters were used to culture SVZ NSCs. BALB/c mice were used as WT mice to match the genetic background of NP-C mice. Further, NPC1^flox/flox or VEGF^flox/flox mice, which are of a C57BL background, were used as the +/+ group (control mouse) to compare with Nestin-cre; NPC1^flox/flox, GFAP-creER; NPC1^flox/flox, CX₃CR1-creER; NPC1^flox/flox, and Nestin-cre; VEGF^flox/flox mice. Both male and female mice were used for all experiments except behavioral studies (which used male mice only). The block randomization method was used to allocate the animals to experimental groups. To eliminate the bias, we were blinded in experimental progress such as data collection and data analysis. Mice were housed at a 12-h day/12-h night cycle with free access to tap water and food pellets. Mouse studies were approved by the Kyungpook National University Institutional Animal Care and Use Committee (IACUC).

Cell Isolation and Culture

Neural progenitors and microglia from the SVZ, thalamus, or cortex were isolated from the mouse brain as previously described.⁴⁷ In brief, the SVZ, thalamus, or cortex of WT (3 or 6- to 8-weeks-old), NP-C (3 or 6- to 8-weeks-old), Nestin-cre; NPC1^flox/flox (6- to 8-weeks-old), Nestin-cre; VEGF^flox/flox (6- to 8-weeks-old), or VEGF^tg (3 weeks old) mice were minced in ice-cold hibernate A/B27/glutamax medium (HABG) (all from Invitrogen) and dissociated using papain (Worthington) solution. After tissue trituration, cells were separated by Optiprep (Sigma-Aldrich) density gradient centrifugation. Fractionated neural progenitors and microglia were cultured in Neurobasal A (Invitrogen)/B27 medium with glutamax (0.5 mM), gentamycin (10 μg/mL, Invitrogen), mouse FGF2 (mFGF2, 5 ng/mL, Invitrogen), and mouse platelet-derived growth factor-bb (mPDGFbb, 5 ng/mL, Invitrogen) or DMEM/nutrient F-12 (DMEM/F12, Sigma-Aldrich) containing 10% fetal bovine serum (FBS, Invitrogen) and 1% penicillin-streptomycin (Invitrogen). Neural progenitors proliferated in suspension and formed aggregates referred to as NSs. Every 2 days, half of the medium was replaced with fresh culture medium.

Primary astrocyte cultures were prepared from C57BL/6 mice as described previously.⁴⁸ In brief, after removal of the meninges, postnatal day 7 (P7) mouse brain tissues were minced and incubated in a rocking water bath at 37°C for 30 min in the presence of 0.25% trypsin-EDTA (Sigma-Aldrich). Enzyme-digested dissociated cells were triturated with astrocyte-specific medium (DMEM/F12 containing 10% FBS, 1% penicillin-streptomycin) and centrifuged at 1,300 rpm for 8 min. The pellet was re-suspended in DMEM/F12 and passed through a 40-μm cell strainer. The filtrate was allowed pre-adherence for 30 min to remove any contamination from fibroblasts before being seeded in dishes and added to astrocyte-specific medium. For astrocyte splitting, dishes were added with cytosine arabinoside (Ara-C) and placed in a heated shaker for 6–7 h. The medium was removed from the dishes, and trypsin-EDTA was added to the dish and incubated for 5–10 min at 37°C. After centrifugation at 1,300 rpm for 8 min, the supernatant was removed and the astrocytes were maintained in culture by feeding every 1 to 2 weeks with astrocyte-specific medium. We confirmed successful isolation of primary cells from mouse brain following each published method.47, 48

MPIO Preparation and Surgical Injection

As a neural progenitor labeling agent, we used 1.63-μm diameter polystyrene- and divinylbenzene-coated fluorescent iron oxide particles (MPIOs, 3.00 mg Fe/mL, green fluorescent dye, 480 nm excitation, 520 nm emission; Bangs Laboratories, Fishers, IN, USA) as previously described.²⁶ For efficient visualization, a non-viral transfection agent was used: poly-L-lysine hydro-bromide (PLL, MW > 300 kDa, Sigma-Aldrich). Prior to injection, the PLL stock solution of 1.5 mg/mL was further diluted to 0.3 mg/mL and mixed with the MPIOs (3.0 mg Fe/mL), yielding a volume of 1.5 μL containing final MPIO concentration of 0.67 mg Fe/mL with 0.045 mg PLL/mL. Transfection agent MPIO complex was placed on a rotating shaker at 600 rpm for 60 min at room temperature to allow the transfection agent to hybridize with the iron oxide particles.

The mice were anesthetized with a combination of 100 mg/kg ketamine and 10 mg/kg xylazine and positioned in a stereotactic head frame (David Kopf Instruments, Tujunga, CA, USA). Stereotactic coordinates were as follows for the LV injection: +0.3 mm anterial, ± 1 mm lateral, −2.3 mm dorsal relative to bregma. A microsyringe (Hamilton) was connected with a pump to inject at a constant rate of 0.50 μL/min. After injection, the needle was left in place for an additional 5 min and then slowly withdrawn. MPIO-injected mice were sacrificed after 1 or 3 weeks to investigate migration of MPIO-labeled neural progenitors using a laser-scanning confocal microscope (FV1000; Olympus) or Slide scan system (Pannoramic scan).

Intraventricular or Thalamic Injections

Three days before the first injection, all mice (3- to 4-weeks-old) underwent surgery for implantation of a guide cannula into its brain. In brief, after anesthesia, a stainless steel cannula was implanted in the LV (LV coordinates described above) or thalamus (−2 mm anterior, ± 1.5 mm lateral, −2.5 mm dorsal relative to bregma) using a stereotaxic frame. To confirm the effect of SVZ degeneration on whole-brain environment, Split-cre plasmid a 1/1.5 CCre/NCre vector ratio was used as previously described.³⁰ NPC1^flox/flox, VEGF^flox/flox, or NPC1^flox/flox; td-tomato^flox/flox mice were daily injected with Split-cre into the LV for 4 weeks. As the control groups in each experiment, 3 μL of negative control vector were implanted. For NSC injection, we dissociated NSs cultured from the SVZ of WT, VEGF^tg, and NP-C mice into single cells and re-suspended the cells. About 3 μL of a suspension of WT, VEGF^tg, or NP-C NSCs in PBS (∼1 × 10⁶ cells) was injected at a rate of 0.3 μL/min for 4 weeks (twice a week) through the cannula. VEGF OX vector also was injected at same condition. The control groups were administered 3 μL of PBS into the LV. WT, VEGF^tg NSCs, and VEGF OX vector were used for intraventricular injection in NP-C mice, and WT NSCs, NP-C NSCs, control shRNA ,or Npc1 shRNA were used for thalamic injection in WT mice. After injection, each mouse was kept in an individual cage to prevent removal of the guide cannula by other mice. The beam, rotarod, open field, hot-plate, and tail-flick tests were sequentially performed one test at a time for 5 days in 8-week-old NP-C mice treated with PBS or VEGF^tg NSCs after 3 days of habituation. 9-week-old mice were sacrificed to investigate brain pathology. All behavioral tests and brain pathology analyses were performed on the same mice in each group. To assess the lifespan, the survival of another group of NP-C mice was weekly checked until mice died together with each treatment. The timeline of the experiment is outlined in Figure 6A. We used the block randomization method to allocate the animals to experimental groups.

NS Formation Assay

To detect neural progenitors in the SVZ, thalamus, or cortex, NSs were cultured as previously described.⁸ NSs were mechanically dissociated, and the resulting viable individual cells were counted. These cells were plated 1 × 10⁴ cells per well in uncoated 24-well plates for NS culture. After 7 days of incubation, formed NSs were counted in each well using a microscope. A minimum cutoff size of 50 μm in diameter was used in defining an NS. All experiments were carried out at least three times, and at least six wells per condition and per experiment were counted. These experiments were repeated six times. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

ELISA

“Off/on” signal, pro-inflammatory cytokines, and VEGF protein levels were assayed by using mouse BDNF (Mybiosource), CD47 (Mybiosource), CD200 (Mybiosource), CXCL10 (R&D System), MMP3 (R&D Systems), IL-1β (R&D Systems), TNF-α (R&D Systems), and VEGF (R&D Systems) according to the manufacturer’s instructions. These experiments were repeated six times. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

VEGF Small Interfering RNA (siRNA) and IL-1β Treatment

Mouse VEGF siRNA (Dharmacon) and scrambled sequence siRNA control (Dharmacon) were treated in astrocytes for 48 h. Cells were further treated with mouse IL-1β (10 ng/mL, R&D Systems) for 24 h following siRNA treatment. In another experiment, astrocytes were respectively treated with IL-1β (10 ng/mL) and IL-1β antibody (10 ng/mL; R&D Systems, AF-401). After 24 h, cells were used for migration assay. These experiments were repeated six times. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

Cell Migration Transwell Assay

The in vitro migration of neural progenitors to reactive astrocytes was detected using a transwell assay.⁴⁹ Each well of a 24-well cell culture plate was separated into two chambers by the insertion of 8-μm pore membranes (Corning, 3422). Astrocytes (1 × 10⁴ cells) treated with CM collected from SVZ-derived microglia of WT, NP-C, control, Nestin-cre; NPC1^flox/flox, or Nestin-cre; VEGF^flox/flox mice or treated with IL-1β (10 ng/mL) were placed into each lower chamber. Neural progenitors (1 × 10⁴ cells) were then seeded into the upper chambers. After 6 h of incubation at 37°C, neural progenitors in the bottom chambers were fixed with 95% ethanol. Non-migrating cells on the upper side of the membrane were discarded. Migrated neural progenitors fixed to the bottom chamber were stained using SOX2 antibodies, and all SOX2-positive cells were quantified under a microscope. These experiments were repeated six times. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

Lentivirus Production

We cloned Npc1 shRNA into lentiviral vector plasmid CS-CDF-CG-PRE. The following short hairpin sequences were used: 5′-AGTTCCAG-TACGGCTCCAA-3′ (SABiosciences-QIAGEN; KM03041N; Npc1-shRNA #3); 5′-GGAATCTCATTCGATGCATAC-3′ (SABiosciences-QIAGEN; negative control shRNA). The Npc1 shRNA were produced by transient transfection of 293T cells.⁵⁰ For production of VEGF OX vector, mouse vascular endothelial growth factor (mVEGF) cDNA was cloned into lentiviral vector plasmid LV-Ubc-MCS-IRES EGFP that was derived from CS-CDF-CG-PRE by replacing cytomegalovirus (CMV) promoter with Ubc promoter and EGFP with IRES EGFP. mVEGF cDNA was amplified from VEGF vector (MGC 12075 clone, Thermo Scientific) and was digested by AgeI-SmaI and ligated to AgeI-SmaI sites of LV-Ubc-MCS-IRES EGFP, resulting in LV-Ubc-mVEGF-IRES EGFP. LV-Ubc-mVEGF-IRES EGFP was also produced by transient transfection of 293T cells.⁵¹ Split-cre was produced as described in previous studies.30, 52 NSC-specific Split-cre was achieved by splitting the Cre-recombinase into two fragments and fusing them to the dimerizing GCN4 coiled-coil domain yielding the N-terminal NCre (hGFAP-NCre) and the C-terminal CCre fragment (Prominin-P2-CCre). Because two different promoters control the expression of the two Cre fragments, only NSCs coexpressing the two markers have functional Cre-recombinase. Lentiviral plasmids, which are consistent with 106-FUV-P2-CCre containing the C-terminal portion of Cre under the Prominin-P2 (CD133) promoter and 109-FUV-hGFAP-NCre containing the N-terminal portion of Cre (fused to GCN4) under the human GFAP promoter, were kindly provided from Prof. Dr. Magdalena Götz (University of Munich, Germany). 106-FUV-P2-CCre and 109-FUV-hGFAP-NCre were transfected in 293T cells for production. Virus-containing media were collected, filtered, and concentrated by ultracentrifugation at 50,000 × g for 2 h and re-suspended in PBS. Viral titers were measured by serial dilution on 293T cells, followed by flow cytometry analysis 48 h later. The titer of the virus used ranged between 2 and 5 × 10⁹ plaque-forming units per mL. 3 μl of Npc1 shRNA was administered into the thalamus of 4-week-old WT mice, and VEGF OX vector was injected into the LV of 4-week-old NP-C mice. Split-cre (1/1.5 CCre/NCre vector ratio) was injected daily into the LV of 4-week-old NPC1^flox/flox, VEGF^flox/flox, or NPC1^flox/flox; td-tomato^flox/flox mice for 4 weeks.

Immunofluorescence Staining

For the immunofluorescence staining, brain was cut on the vibratome (30 μm). The following antibodies were used: Iba1 (rabbit, 1:500; Wako, 019-19941), GFAP (rabbit, 1:500; Dako, N1506), GFAP (chicken, 1:500; Abcam, AB4674), SOX2 (mouse, 1:100; R&D Systems, MAB2018), MASH1 (mouse, 1:100; BD Biosciences, 556604), Doublecortin (DCX, rabbit, 1:400; Abcam, AB18723), VEGF (rabbit, 1:100; Abcam, AB39250), and NeuN (mouse, 1:200; Millipore, MAB377). All were visualized using Alexa anti-mouse 488 and 594, Alexa anti-rabbit 488 and 594, and anti-chicken 488 as secondary antibodies. For filipin staining, brain sections were fixed with 4% paraformaldehyde for 15 min, washed with PBS, and incubated at room temperature for 30 min in the dark with a staining solution containing 100 μg/mL filipin (Polysciences) in PBS. The sections were analyzed with a laser-scanning confocal microscope (FV1000; Olympus) or with a BX51 microscope (Olympus). SVZ cells and neurons were quantified using stereology. For quantification of SVZ cells, systematic random sampling was obtained using every fifth in a series of 30-μm-thick coronal sections through the striatal SVZ. The equidistant sections (five per brain) were used for stereological quantification to assess individual cells in the SVZ. All counts were performed through a 60× objective lens on an Olympus BX51 microscope equipped with a DP70 digital camera, and the Visiomorph software (Visiomorph) applying a 120 × 120 μm sampling grid and 25 × 25 μm counting frame, allowing the discrimination of individual cells. NeuN-positive cells were counted using a 100× objective and counted as neurons. Sampling schemes in the thalamus (1:6 series) were as follows: LGNd,125 × 125 μm sampling grid, 74 × 42 μm counting frame; VPM, 175 × 175 μm sampling grid, 74 × 42 μm counting frame; VPL, 100 × 100 μm sampling grid, 74 × 42 μm counting frame. Sampling schemes in the S1BF (1:12 series) were as follows: lamina IV, 150 × 150 μm sampling grid, 41 × 26 μm counting frame; lamina VI, 200 × 200 μm sampling grid, 60 × 40 μm counting frame. MetaMorph software (Molecular Devices) was used to calculate the average intensity. IMARIS software (Bitplane)¹⁷ was used for analysis of three-dimensional reconstruction of microglia and astrocytes. Confocal images were taken through a z stack (total z axis length = 10 μm) and were imported into the IMARIS software. Cell body width was measured, and cell dendrites were automatically detected using the analysis tool. Then the image was converted into 3D image, and the cell body volume, process length, number of branches, and terminal tips were automatically quantified. Data were collected from 4–6 mice per group. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

Flow Cytometry

Preparation of single-cell suspension and flow cytometry for analysis of adult astrocytes, were done based on earlier protocols.⁵³ The general procedure was as follows: the mice were anesthetized and perfused with Hank’s Balanced Salt Solution without calcium or magnesium (HBSS) (Invitrogen). The cortex or SVZ were dissected out and kept in ice-cold HBSS. After mechanical dissociation, the tissue was subjected to enzymatic dissociation using papain solution and incubates at 37°C for 50 min on a rocker. The mix was spun at 200 × g for 15 min, and the pellet was triturated to obtain a single-cell suspension. After enzymatic dissociation, cells were re-suspended in 30% Percoll (GE Healthcare) and centrifuged for 10 min at 700 × g. The supernatant containing the myelin was removed, and the pelleted cells were washed with HBSS. For staining of astrocytes, pelleted cells were re-suspended in PBS (0.5% BSA) and stained with PE-GLAST-1 (Miltenyi Biotec, 130-095-821), APC-CD11b (eBioscience, 17-0112-82), or VEGF (Abcam, ab52917) at 4°C in the dark for 30 min. Alexa anti-rabbit 488 (Invitrogen, A11008) was used as secondary antibody for staining VEGF antibody. GLAST-1⁺CD11b⁻ cell population were sorted as astrocytes using AriaIII (BD Biociences) and used for RNA or lipid extraction. For analysis of VEGF derived from neural progenitors in the brain and neural progenitors or neurons in the OB or thalamus, single cells were prepared as described above. Single cells were stained with VEGF (Abcam, ab52917), DCX (Abcam, AB18723), or NeuN (Millipore, MAB377) and then further stained with Alexa anti-rabbit 594 (Invitrogen, R37117) or Alexa anti-mouse 633 (Invitrogen, A21052). Flow cytometry data were further analyzed using FlowJo software (Tree Star). Data were collected from six mice per group. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

Lipid Extraction and Sphingosine/Sphingomyelin Quantification

The cortex or cells were lysed in homogenization buffer containing 50 mM HEPES (Gibco), 150 mM NaCl (Sigma-Aldrich), 0.2% Igepal (Sigma-Aldrich), and protease inhibitor (Calbiochem).²¹ To quantify the sphingosine and sphingomyelin levels, the dried lipid extract was re-suspended in 0.2% Igepal CA-630. Four microliters of the lipid extracts was added into 20 mL of naphthalene-2,3-dicarboxaldehyde (NDA) derivatization reaction mixture (25 mM borate buffer, pH 9.0, containing 2.5 mM each of NDA and NaCN). The reaction mixture was diluted 1:3 with ethanol, incubated at 50°C for 10 min, and centrifuged (13,000 × g for 5 min). An aliquot (30 μL) of the supernatant was then transferred to a sampling glass vial, and 5 μL was applied onto an ultra-performance liquid chromatography (UPLC) system for analysis. The fluorescent sphingosine derivatives were monitored using a model 474 scanning fluorescence detector (Waters). Quantification of the sphingosine and sphingomyelin peaks were calculated from sphingosine and sphingomyelin standard calibration curves using the Waters Millennium software. Data were collected from 4 to 6 mice per group. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

Amplex Red Assay

The cells and brain tissue were lysed with lysis buffer (50 mM phosphate buffer, 500 mM NaCl, 25 mM cholic acid, and 0.5% Triron X-100). The unesterified cholesterol was determined using the Amplex Red Cholesterol Assay Kit (Molecular Probes) according to the manufacturer’s instructions. After incubation for 30 min at 37°C, the fluorescence intensities were measured on a fluorescence microplate reader (Molecular Devices) equipped with a filter set for excitation and emission at 560 ± 10 nm and 590 ± 10 nm, respectively. The cholesterol content was calculated with a cholesterol standard curve. Cellular cholesterol content was normalized to protein content. Data were collected from 4 to 6 mice per group. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded.

Drug Treatments

To compare synergistic effects between VEGF^tg NSCs and ibuprofen when combined with CD, mice were given a subcutaneous injection of CD (4000 mg/kg; H107, Sigma Aldrich) starting at 1 week old (once a week), intraventricular injection of VEGF^tg NSCs starting at 4 weeks old (twice a week), or ibuprofen (100 mg/kg/day; I4883, Sigma Aldrich) as a dry admixture to mouse chow starting at 6 weeks old due to the toxicity seen with earlier dosing. Body weight and rotarod tests were measured every week, and then mice were sacrificed at 10 weeks old for analysis. To assess the lifespan, the survival of another group of NP-C mice was weekly checked until mice died together with each treatment. The timeline of the experiment is outlined in Figure 8A.

Behavioral Studies

To assess motor function behavior, open-field, rotarod, and beam tests were performed. The open-field box consisted of a square box. Each animal was placed in the box for 10 min. Overall activity in the box was measured, and the amount of time and distance traveled in the center arena was noted. After each trial, the test chambers were cleaned with a damp towel and distilled water followed by 70% alcohol. Rotarod apparatus (accelerating model 47600; Ugo Basile) was set to an initial speed of 4 rpm, and the acceleration was increased by 32 rpm every 25 to 30 s. Scores were registered every 2 days, and three independent tests were performed at each measurement. Uniform conditions were carefully maintained for each test, and there was a rest time of 1 h between trials. Each test was limited to 300 s. Balancing and hindlimb placement were evaluated by assessing the ability of mice to traverse two types of balance beams to reach a safety platform. Each mouse was tested for its ability to traverse two different styles of 41-cm-long scored Plexiglas beams. One was 12 mm in diameter, and the other was 6 mm wide. Beams were placed horizontally 50 cm above a table. Time to traverse each beam was recorded for each trial with a 60 s maximum cutoff, and falls were scored as 60 s.

Hot-plate and tail-flick test were performed to assess sensory function. For hot-plate test, the mice were placed on a heating apparatus (Panlab/Harvard Apparatus, Spain) maintained at 50°C. The time that elapsed until the mouse licked its hindpaw or jumped was measured by an experienced observer. The test was repeated three times at an interval of 15 min, and the median values were analyzed. For tail-flick test, each mouse was wrapped with black velvet, and its tail was heated (Panlab/Harvard Apparatus, Spain). The latency to flick the tail was measured automatically. The median values of three measurements were collected. This experiment was performed in a blind fashion, and data were collected from 10–13 mice per group. Randomization procedures are not applicable to these experiments. G-power software was used for sample size estimation. No data were excluded except when the mice were dead.

RNA Isolation and Real-Time PCR Analysis

RNA was extracted from the brain homogenates and cell lysates using the RNeasy lipid tissue mini kit and RNeasy Plus mini kit (QIAGEN) according to the manufacturer’s instructions. cDNA was synthesized from 5 μg of total RNA using a commercially available kit (Takara Bio). Quantitative real-time PCR was performed using a Corbett research RG-6000 real-time PCR instrument. Used primers are described in Table S1.

Statistical Analysis

The numbers of samples used to assess significant differences were calculated using G-power software with a significance level of 0.05 and a power of 0.8, and exact sample size was indicated within each figure legend. Comparisons between two groups were performed with the Student’s t test. In cases where more than two groups were compared to each other, a one-way ANOVA was used, followed by Tukey’s honestly significant difference (HSD) test. Statistical differences among survival curves were determined using a log rank test. All statistical analyses were performed using Prism 7 software. *p < 0.05, **p < 0.01, ***p < 0.001 were considered to be significant.

Author Contributions

M.H.P. designed and performed experiments and wrote the paper. B.J.C., M.S.J., J.Y.L., I.K.J., K.H.P., H.W.L., and T.Y. performed experiments and analyzed data. B.H.L., E.H.S., H.K.J., and J.-s.B. interpreted the data and reviewed the paper. H.K.J. and J.-s.B. designed the study and wrote the paper. H.H.M. generated and provided VEGF^tg mice. All authors discussed results and commented on the manuscript.

Conflicts of Interest

The authors declare no conflict of interest.