A total of 80 mice were used in this study. Blood clots were accumulated mainly around the circle of Willis and the ventral surface of brainstem after SAH induction, no blood clot was observed in the sham group (Additional file 1: Fig. S1). SAH caused an overall mortality rate of 8.75% (7/80), and no mouse died in the sham group. PI staining showed that SAH induced significant neuronal death and the PI-positive cells mainly distributed around the SGZ (Additional file 1: Fig. S2). Immunofluorescent images revealed that SAH significantly induced astrocyte reactivation in the hippocampus at 3 days after SAH, as indicated by the large increase in GFAP immunoreactivity (Additional file 1: Fig. S3).
To target the reactive astrocytes, astrocytic promoter GFAP was used to drive the expression of NeuroD1 and reporter GFP in astrocytes specifically. ND1–GFP or GFP virus was injected into the hippocampus at 3 days after SAH induction, and then the relative tissue was harvested at 3 dpi, 7 dpi, 14 dpi as illustrated in Additional file 1: Fig. S4A. Immunofluorescent images showed that ND1–GFP efficiently increased the expression of NeuroD1 in the hippocampus compared with that in the GFP group at each relative timepoint (Additional file 1: Fig. S4B–D).
NeuroD1-attenuated reactive astrocyte-mediated neuroinflammation after SAH
Reactive astrocyte acts as a two-edge sword which may induce harmful effect by secreting inflammatory cytokines or serve as a physical barrier to prevent spreading of injury. We wondered what impact may occur to reactive astrocyte after NeuroD1 treatment. As illustrated in Additional file 1: Fig. S5A, the immunofluorescent images revealed that NeuroD1 treatment significantly ameliorated the activation of astrocyte in the hippocampus than that in the GFP group at 3dpi, 7dpi and 14dpi (Additional file 1: Fig. S5B–D). C3 was used to label neurotoxic A1 astrocytes, and immunofluorescent staining showed that NeuroD1 significantly decreased the number of C3-labeled GFAP-positive astrocytes than that in the GFP group at 14 dpi (Fig. 1A). Representative images of reactive astrocytes in the hippocampus were traced into outline and branch patterns, and the quantitative analysis showed that the branch count, cellular perimeter, soma volume and sum dendritic length of the reactive astrocytes were all significantly increased after SAH. However, NeuroD1 treatment markedly reversed all these parameters than those in the GFP group at 14 dpi (P < 0.05, Fig. 1B–E.
PTX3 was introduced as a molecular marker for neuroprotective A2 astrocytes, and western blot showed that PTX3 was significantly decreased after SAH. However, NeuroD1 treatment promisingly increased the protein level of PTX3 than that in the GFP group (P < 0.05, Fig. 1F, H). Furthermore, western blot showed that C3 was significantly increased after SAH, and NeuroD1 markedly decreased the C3 protein level compared with that in the GFP group at 14 dpi (P < 0.05, Fig. 1F, G).
Astrocyte can function by interacting with blood vessel and contributing to BBB to limit harmful stimulation from chemical and biological toxicity. SAH can induce severe BBB impairment as reported in our previous studies[1, 22]. In this study, immunohistochemistry for LY6C and quantification analysis showed that SAH caused significant blood vessel swollen at 14 dpi, ND1–GFP treatment markedly reversed the hypertrophic blood vessels than those in the GFP group (Fg. 1I, J), which indicated that ND1–GFP treatment restored the leaky BBB following SAH.
A1 astrocyte can mediate neuroinflammation by activating neurotoxic microglia and releasing inflammatory cytokines . iNOS was used to mark harmful microglia, and immunofluorescent results showed that injection of ND1–GFP markedly reduced the number of iNOS+Iba-1+ microglia than that in the GFP group (Fig. 2A). Notably, microglia contacting NeuroD1-infected astrocytes exhibited less amoeboid shape and more processes as compared to the microglia contacting AAV–GFP-infected astrocytes (Fig. 2A). Moreover, immunofluorescent images showed that ND1–GFP administration significantly reduced the amount of C1q than that in the GFP group (Fig. 2B). Furthermore, western blot showed that inflammatory cytokines iNOS, TNF-α and IL-18 were significantly increased after SAH induction, and NeuroD1 treatment strongly reduced the amount of iNOS, TNF-α and IL-18 than those in the GFP group (P < 0.05, Fig. 2C–F). Taken together, these results indicated that NeuroD1-attenuated reactive astrocyte-mediated neuroinflammation following SAH.
NeuroD1 improved the microenvironment and boosted neurogenesis following SAH
The SGZ is one of the most important NSC niches which is vulnerable to the neuroinflammatory microenvironment . To determine the impact of SAH and subsequent NeuroD1 application on NSC niches, we tested Nestin, a molecular marker for NSC. ND1–GFP virus was injected to the left hippocampus, and the vehicle was injected to the right hippocampus of the same mouse at 3 days after SAH induction (Fig. 3A). Immunofluorescent images showed that the expression of Nestin was significantly higher in the ND1–GFP-injected side than that in the vehicle side at 14 dpi (Fig. 3B). Interestingly, ND1–GFP promisingly increased the number of Ki-67 positive cells over the control side, and Ki-67 positive cells were in a region with relatively higher Nestin expression levels (Fig. 3C). Furthermore, we conducted experiments that injected ND1–GFP virus and control GFP virus using separate mice, and the immunofluorescent images showed that ND1–GFP treatment significantly upregulated the expression of Nestin than that in the GFP group at 14 dpi (Fig. 3D). Moreover, SOX2, another molecular marker for NSC, was used to identify the cellular localization of Ki-67, and immunofluorescent images showed that ND1–GFP markedly increased the number of SOX2+Ki-67+ cells in the hippocampus than those in the GFP group at 14 dpi (Fig. 3E), suggesting that ND1–GFP altered the cellular microenvironment favoring the proliferation of NSCs in SGZ at 14 dpi.
Astrocyte-derived PTN can reduce pro-inflammatory signaling and is crucial for the maturation of newborn neurons [25, 26]. We wondered whether NeuroD1 treatment affected the secretion of PTN, and western blot revealed that the protein level of PTN was significantly higher in the ND1–GFP group than that in the GFP group at 14 dpi (P < 0.05, Fig. 3F, G). Furthermore, immunofluorescent images also showed that ND1–GFP injection significantly increased the number of DCX-positive neurons at SGZ than that in the GFP group at 14 dpi (Fig. 3H). Together, these results indicated that ND1–GFP treatment-boosted neurogenesis, which likely benefited from the improvement of neuroinflammatory microenvironment following SAH.
NeuroD1-converted transfected cells into neurons at the early phase and boosted endogenous neurogenesis at the late phase of SAH
Whether NeuroD1 can induce astrocytes to neurons conversion in the hippocampus after SAH remains unclear. Immunofluorescent staining was conducted, and the results showed that ND1–GFP significantly increased the density of GFP+DCX+ cells than that in the GFP group at 14 dpi (Fig. 4A). Furthermore, mature neuronal marker NeuN was employed, and the immunofluorescent images showed that NeuroD1 induced significant conversion of GFAP+GFP+ into NeuN+GFP+ cells at both 7 dpi and 14 dpi (Fig. 4B), while few NeuN+GFP+ cells were identified in the GFP group. Interestingly, some ND1–GFP-infected cells were positive for both GFAP (magenta) and NeuN (red) as shown by the white arrowhead in Fig. 4B, which indicated a transitional stage during astrocytes to neurons conversion. Western blot again showed that the protein levels of both DCX and NeuN were significantly higher in the ND1–GFP group than that in the GFP group (P < 0.05, Fig. 4C–E). Collectively, these data indicated that NeuroD1 efficiently converted transfected cells, most likely astrocytes, into neurons at the early phase of SAH.
To understand how NeuroD1 impacted neurogenesis, we harvested the brain at 14 dpi and at 23 dpi, as illustrated in Fig. 5A. Compared to the control group, ND1–GFP significantly increased the expression of NeuroD1 throughout the hippocampus at 14 dpi and at 23 dpi (Fig. 5B, C). Again, the results showed that NeuroD1 significantly increased the number of DCX-labeled newborn neurons than that in the contralateral control side both at 14 dpi and at 23 dpi (Fig. 3B, C). Interestingly, at 14 dpi, immunofluorescent images showed that ectopic NeuroD1 mainly promoted the expression of DCX in area away from SGZ known for endogenous NSCs (Fig. 5B), indicating that DCX-labeled cells arose mainly from the in vivo reprogramming, but not endogenous neurogenesis. However, at 23 dpi, the expression of DCX mainly distributed at the region around SGZ (Fig. 5C), suggesting that NeuroD1 mainly boosted endogenous neurogenesis at the late phase of SAH. These results suggested that ectopic NeuroD1 mainly mediated astrocyte reprogramming at the early phase of SAH and restored endogenous neurogenesis at the late phase of SAH.
NeuroD1 repaired the neural circuit and improved neurocognitive function after SAH
Neuron to neuron synaptic connections form the neural circuit and are the cellular basis for brain function. Dendritic spine density is a representative for synaptic plasticity, which is closely related to learning and memory functions . In this study, we found that SAH caused significant reduction in dendritic spine protrusion length and spine density. However, NeuroD1 treatment dramatically increased the dendritic spine protrusion length and the spine density than those in the GFP group (P < 0.05, Fig. 6A–C). Furthermore, the protein levels of the pre-synaptic marker Syn and the post-synaptic marker PSD95 were measured to assess the strength of synaptic connection in the hippocampus after SAH, immunofluorescent images showed that ND1–GFP treatment significantly increased the protein levels of both PSD95 and Syn than those in the GFP group (Fig. 6D), and western blot revealed that the protein levels of PSD95 and Syn were both significantly decreased after SAH induction (P < 0.05, Fig. 6E–G). However, NeuroD1 injection strongly increased the protein level of Syn than that in the GFP group, though the difference did not reach significance (P = 0.17, Fig. 6E–G). Moreover, NeuroD1 significantly increased the expression of PSD95 than that in the GFP group (P < 0.05, Fig. 6E–G), indicating that NeuroD1-induced neurogenesis significantly repaired the disrupted synapses after SAH.
To further determine whether the repaired synaptic connections by NeuroD1 was functional, we used neurophysiological recording to evaluate neural circuit function. The firing rate of neurons in the hippocampus were markedly impaired following SAH. However, NeuroD1 significantly increased the firing rate of neurons than that in the GFP group (Fig. 6H). Moreover, SAH induced dramatical decrease of the number of neuronal discharges, and NeuroD1 treatment significantly increased the number of neuronal discharge than that in the GFP group (Fig. 6I). Together, NeuroD1 efficiently repaired the neural circuit through mediating astrocyte to neuron conversion and restoring endogenous neurogenesis after SAH.
Novel object recognition and Morris water maze were performed to assess the role of NeuroD1 on functional recovery following SAH. Novel object recognition was conducted at 11–14 dpi, and the results indicated that novel object recognition discrimination index was impaired after SAH, and NeuroD1 treatment significantly rescued the results than those in the GFP group (P < 0.05, Fig. 7A). Morris water maze was assessed at 9–14 dpi and the results showed that the number of platform crossing was significantly decreased and the distance to platform was markedly increased after SAH induction (P < 0.05, Fig. 7B–D. However, NeuroD1 treatment significantly increased the number of platform crossings than those in the GFP group, and the distance to platform was significantly shortened in the NeuroD1 group than that in the control group (P < 0.05, Fig. 7C–D, suggesting that injection of NeuroD1 markedly improved the neurocognitive function after SAH. After removal of platform at 14 dpi, swimming path tracks and heat maps showed that the memory of the mouse was significantly improved in the NeuroD1 group than that in the GFP group (P < 0.05, Fig. 7E, F).