In vivo neural regeneration via AAV-NeuroD1 gene delivery to astrocytes in neonatal hypoxic-ischemic brain injury

Background Neonatal hypoxic-ischemic brain injury (HIBI) is a significant contributor to neonatal mortality and long-term neurodevelopmental disability, characterized by massive neuronal loss and reactive astrogliosis. Current therapeutic approaches for neonatal HIBI have been limited to general supportive therapy because of the lack of methods to compensate for irreversible neuronal loss. This study aimed to establish a feasible regenerative therapy for neonatal HIBI utilizing in vivo direct neuronal reprogramming technology. Methods Neonatal HIBI was induced in ICR mice at postnatal day 7 by permanent right common carotid artery occlusion and exposure to hypoxia with 8% oxygen and 92% nitrogen for 90 min. Three days after the injury, NeuroD1 was delivered to reactive astrocytes of the injury site using the astrocyte-tropic adeno-associated viral (AAV) vector AAVShH19. AAVShH19 was engineered with the Cre-FLEX system for long-term tracking of infected cells. Results AAVShH19-mediated ectopic NeuroD1 expression effectively converted astrocytes into GABAergic neurons, and the converted cells exhibited electrophysiological properties and synaptic transmitters. Additionally, we found that NeuroD1-mediated in vivo direct neuronal reprogramming protected injured host neurons and altered the host environment, i.e., decreased the numbers of activated microglia, reactive astrocytes, and toxic A1-type astrocytes, and decreased the expression of pro-inflammatory factors. Furthermore, NeuroD1-treated mice exhibited significantly improved motor functions. Conclusions This study demonstrates that NeuroD1-mediated in vivo direct neuronal reprogramming technology through AAV gene delivery can be a novel regenerative therapy for neonatal HIBI. Supplementary Information The online version contains supplementary material available at 10.1186/s41232-024-00349-y.


Introduction
Neonatal hypoxic-ischemic brain injury (HIBI) accounts for most neonatal central nervous system injuries and leads to neonatal death and long-term neurodevelopmental deficits, including cognitive and learning disabilities, cerebral palsy, and epilepsy [1].Despite the introduction of therapeutic hypothermia and advancements in supportive care, optimal therapeutics to improve the neurological outcomes of HIBI patients are still lacking [2].Of note, the regeneration of functional neurons to replace damaged ones remains an important challenge for clinicians.
In the acute phase of HIBI, astrocytes proliferate and migrate to the lesion sites, forming a barrier for toxic substances.However, once activated, reactive glial cells remain and accumulate at the lesion site, forming a glial scar and secreting neuroinhibitory factors that block axon regeneration [3].Therefore, the accumulation of reactive astrocytes during the later stages after HIBI is considered an obstacle to regenerative therapies, including both endogenous neuroregeneration or exogenous cell transplantation.Recently, in vivo direct neuronal reprogramming technology has emerged as a promising approach for replenishing lost neurons without the need for exogenous cell supply or immunosuppression [4,5].Furthermore, ectopic expression of regulatory transcription factors (e.g., Neurogenin2, Sox2, Ascl1, and neurogenic differentiation factor 1 (NeuroD1)) convert glial cells into functional neurons after brain or spinal cord injury [6][7][8][9][10].
In this study, we established a neuroregenerative strategy for treating neonatal HIBI through astrocyte-to-neuron conversion by ectopic NeuroD1 expression.NeuroD1 was delivered to reactive astrocytes of HIBI mice, using the astrocyte-tropic adeno-associated viral (AAV) vector AAVShH19, engineered with the Cre-FLEX system.We investigated the efficiency of in vivo astrocyte-to-neuron conversion and verified the functional status of neurons in terms of electrophysiological activities and synaptic connections with existing host neurons.Changes in the microenvironment of HIBI and recovery from neurobehavioral deficits were also analyzed.

Primary astrocyte isolation and culture
Primary astrocytes were prepared from the ICR mouse cortex on postnatal day 3 using a previously published protocol [11].Briefly, meninges were removed, the cortex was carefully dissected, and the tissue was digested with trypsin-ethylene-diamine-tetraacetic acid (trypsin-EDTA; Gibco, Carlsbad, CA, USA) for 30 min at 37 °C.After centrifugation (300 g, 5 min), the cell pellet was suspended in astrocyte culture medium (Dulbecco's modified Eagle's medium; Gibco) containing 10% fetal bovine serum (FBS; Gibco) and 0.1% penicillin/streptomycin (Gibco)).These mixed glia cells were plated in a poly-D-lysine-coated T75 tissue culture flask, and the medium was changed to a fresh medium every 2-3 days.After 1 week, microglia were separated from astrocyte monolayer sheets by shaking the flask at 180 rpm for 30 min and then discarding the supernatant.Next, 20 mL of fresh medium was added, the flask was shaken at 240 rpm for 6 h, and the supernatant was discarded to remove oligodendrocytes.The remaining astrocyte layer was rinsed with phosphate-buffered saline (PBS; Promega, Madison, WI, USA).After aspiration of PBS, 5 mL of trypsin-EDTA (Gibco) was added at 37 °C for 10 min, then 5 mL of fresh medium was added, and the cells were spun at 180 g for 5 min.The supernatant was discarded, and 40 mL of fresh medium was added and replaced every 2-3 days.These primary cultured astrocytes were used for neuronal conversion.

AAV vector transgene construction
To construct pAAVShH19-CMV-NeuroD1-T2A-GFP, the NeuroD1, T2A, and GFP genes were inserted into the pAAVShH19-Multi-cloning site (MCS) (Cell Biolab #VPK-410, San Diego, CA, USA) between EcoRI and HindIII using Gibson assembly cloning kit (New England Biolabs, Ipswich, MA, USA) according to the manufacturer's protocol.To construct the pAAVShH19-hGFAP-Cre vector, the hGFAP promoter and the Cre gene were obtained from the hGFAP-Cre plasmid (RRID:Addgene_40591, Watertown, MA, USA) and inserted into pAAVShH19-MCS between MluI and EcoRI to replace the CMV promoter and human β-globin intron.To construct pAAVShH19-FLEX-NeuroD1-T2A-GFP and pAAVShH19-GFP-T2A-GFP, pAAVShH19-FLEX-GFP (RRID:Addgene_28304) was used as the backbone.The NeuroD1, T2A, and GFP genes were inserted into the backbone between KpnI and XhoI using the Gibson assembly cloning kit (New England Biolabs) according to the manufacturer's protocol.

AAV vector production
To produce AAV vectors, HEK293T cells derived from human embryonic kidney cells (ATCC, Manassas, VA, USA) were cultured in Dulbecco's modified Eagle's medium (Corning, Glendale, AZ, USA) supplemented with 10% FBS (Corning) and 1% penicillin/streptomycin (Gibco) at 37 °C under 5% CO 2 .The recombinant AAVshH19 vector, which was engineered to enhance its affinity to glial cells in a previous study [12], was produced in HEK293T cells using a calcium-phosphate transient transfection method as previously described [13].Equal quantities of three plasmids (17 μg each), including the one with a rep/AAVShH19 cap gene (a kind gift from Dr. David Schaffer, University of California, Berkeley), a transgene, and an adenoviral helper gene (pHelper), were transfected into HEK293T cells by complexing with a calcium-phosphate solution.Cells were harvested 72 h post-transfection and lysed by three cycles of freezing and thawing.AAVShH19 crude lysate was treated with a benzonase nuclease (Sigma, St. Louis, MO, USA) and purified by centrifugation at 42,000 rpm for 2 h in discontinuous iodixanol density gradients (Axis-Shield, Dundee, Scotland, UK) with a VTi 65•2 rotor (Beckman Coulter, Brea, CA, USA).The virus-containing layer was extracted, and the buffer exchange was conducted with PBS (Biosesang, Seongnam, Korea) containing 0.01% (v/v) Tween-20 (Sigma) using 100 K MWCO Amicon tubes (Millipore, Billerica, MA, USA) according to the manufacturer's instructions.DNase-resistant viral genome was extracted by Proteinase K (both from Thermo Fisher Scientific, Waltham, MA, USA) and then quantified by quantitative PCR (Mini Opticon; Bio-Rad, Hercules, CA, USA) using the SYBR Green master mix (Thermo Fisher Scientific).

Induction of HIBI and virus injection
The Institutional Animal Care and Use Committee at Yonsei University College of Medicine, Seoul, Korea (permit number: 2019-0011) approved all procedures, which were conducted in accordance with the guidelines of the National Institutes of Health.On postnatal day 7, ICR mice were anesthetized with isoflurane and underwent permanent occlusion of the right common carotid artery.Following the procedure, the pups were kept warm until regaining consciousness and were returned to their dams for 2 h.After this, the pups were exposed to a hypoxic atmosphere consisting of 8% oxygen and 92% nitrogen for a duration of 1.5 h.The body temperature of the mice was maintained at 37 °C using a warm pad.All mice received the same housing and care, including a 12:12 h light-todark cycle, and were evenly distributed among the vehicle, control, and NeuroD1 groups.On day 3 after inducing HIBI, the mice were randomly divided into three groups for AAVShH19 or vehicle injection: (i) PBS (vehicle, 6 µL); (ii) AAV-GFAP::Cre (1.5 × 10 10 vg/µL) together with AAV-CAG::FLEX-GFP-T2A-GFP (1.5 × 10 9 vg/µL; control AAV-GFP, 6 µL); (iii) AAV-GFAP::Cre (1.5 × 10 10 vg/ µL) together with AAV-CAG::FLEX-NeuroD1-T2A-GFP (1.5 × 10 9 vg/µL; AAV-NeuroD1-GFP, 6 µL).The mice were anesthetized and then dissected through the dorsal midline of the scalp.A glass micropipette (0.3 mm in diameter) was used to inject AAV-GFAP::Cre + AAV-CAG::FLEX-GFP, AAV-GFAP::Cre + AAV-CAG::FLEX-NeuroD1-T2A-GFP, or an equivalent amount of vehicle directly into the center of the infarct site.The investigator was blinded to the group allocation of the animals during the experiment.No statistical method was used to predetermine the sample size.The sample size of each experiment is shown in the figure legends.No data were excluded from the analysis.

RNA extraction from brain tissues of the HIBI model
Total RNA was extracted from tissue dissected from the ipsilateral hemisphere with TRIzol reagent (Molecular Research Center, Cincinnati, OH, USA) according to the manufacturer's instructions.The purity of the RNA was evaluated by the A260/A280 ratio, and the RNA concentration was measured using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA).The extracted RNA samples were divided into aliquots and stored at − 80 °C until further use.

Quantitative real-time polymerase chain reaction (qRT-PCR)
To perform gene expression analysis, first-strand cDNA was synthesized using 1 μg of total RNA and a first-strand cDNA synthesis kit (Promega) following the manufacturer's protocol.The cDNA was then amplified using a LightCycler 480 SYBR Green I Master mix (Roche, Mannheim, Germany) in a LightCycler 480 System real-time qPCR machine (Roche).The internal control gene used was Gapdh.The amplification product's specificity was confirmed by melting curve analysis.The following primer pairs were used, with the sense primer listed first followed by the antisense primer:
Whole-cell patch-clamp recordings were obtained from healthy-appearing GFP + cells of the ipsilateral hemisphere.Recording pipettes with 3-6 MΩ tip resistance was made using a micropipette puller (Narishige, PC-100, Amityville, NY, USA) and filled with internal solution containing (in mM) 135 K-gluconate, 10 HEPES, 8 NaCl, 5 Phosphocreatine di (Tris), 4 Mg-ATP, 0.4 Na 2 GTP, and 0.5 EGTA, pH 7.35 with KOH and 285 mOsm.Black-and-white images of the recorded neuron under infrared and fluorescence illumination were captured (HCI live Image, Hamamatsu Photonics, Shizuoka, Japan).Recording cells were clamped with 500ms current pulses from − 100 to 350 pA amplitude with 50 pA increments, and corresponding voltage responses were recorded with a 100 kHz sampling rate (pCLAMP 11, Axon Instruments, San Jose, CA, USA).

Behavioral tests and analysis/behavioral assessment Adhesive removal test
The adhesive removal test was performed to evaluate somatosensory and motor functions and deficits according to a previously described procedure [14].A mouse was placed in a test box (150 cm 2 ) for 1 min to habituate.Thereafter, the mouse was taken out of the box, and two small pieces of adhesive tape (0.3 × 0.3 cm) were attached with equal pressure to each forepaw.The order of tape application (right or left) was alternated between each animal.The mouse was then placed in a test box, and the time to first contact and removal from each paw was recorded.A test time of up to 120 s was allowed for mice to remove the adhesive tape.The adhesive removal test was performed 2, 3, and 4 weeks after virus injection.

Rotarod test
To evaluate motor coordination and balance, the rotarod test was conducted using a knurled white plastic rod with a diameter of 3 cm and a width of 6 cm (LSi Letica; Panlab, Barcelona, Spain).The mice were placed on the rotating rod (4 rpm) in the opposite direction of their movement to become familiar with the setup.The rotational speed was gradually increased up to 40 rpm and then maintained at this final speed for 5 min; the latency to fall from the rod was recorded.The experiment was terminated when the animal fell off the rung or grabbed the apparatus and rotated twice in succession.The average of three consecutive trials was considered as the latency to fall off the rod.A researcher who was blinded to the treatment groups randomized all neonatal HIinjured mice into three experimental groups (vehicle, control, and NeuroD1).

Statistics
All cell experiments were performed in triplicate.No statistical methods were used to predetermine the sample size.The Statistical Package for Social Sciences (SPSS) software (version 25; IBM Corporation, Armonk, NY, USA) was used for statistical analysis.Data are represented as mean ± standard error of the mean (SEM).One-way analysis of variance was used to compare multiple groups, followed by Bonferroni post hoc analysis for pairwise comparisons of groups.A Student's t test (twotailed unpaired) was used when comparing two values.p < 0.05 indicated a statistically significant difference.
Next, to investigate which types of neurons were differentiated from astrocytes, we stained for Glutamic Acid Decarboxylase 67 (GAD67) and Gamma-aminobutyric acid (GABA; GABAergic neurons); Vesicular glutamate transporter 1 (VGluT1; glutamatergic neurons); and Choline acetyltransferase (ChAT; cholinergic neurons).NeuroD1-infected astrocytes were immunopositive for GAD67 and GABA but not VGluT1 or ChAT (Fig. 1E,  I-K).These results demonstrate that NeuroD1 can directly induce the conversion of mouse astrocytes into predominantly GABAergic neurons.
To assess synaptic connectivity, we stained for synaptophysin to visualize synaptic vesicles.Synaptophysin was observed in converted cells at 60 dpi (Fig. 6A,  B), indicating the presence of synaptic structures.To further investigate whether NeuroD1-converted neurons have electrophysiological characteristics of functional excitable neurons, we measured the relationship between stimulating inputs and action potential (AP) firing frequency in NeuroD1-converted neurons using cortical slice whole-cell patch-clamp recordings in current clamp mode at 60 dpi.Cortical slice recordings were performed on GFP-labeled cells of the AAV-Neu-roD1 and AAV-GFP groups (Fig. 6C-E).GFP + cells in the AAV-NeuroD1 group exhibited repetitive APs, their firing frequency increased as higher amplitudes of current were applied, ranging from 50 to 350 pA (Fig. 6D).By contrast, cells in the AAV-GFP group showed little or no AP firing across the entire range of current amplitude (Fig. 6E).

Astrocyte-to-neuron conversion attenuates microglial inflammatory responses and reduces glial scars after HIBI
As microglia activation contributes to inflammation in HIBI, we evaluated the effect of in vivo direct neuronal reprogramming on microglia at 30 dpi.Immunostaining with Iba1, a marker for microglia and macrophages, revealed a significant increase in AAV-GFP-infected areas after HIBI, whereas AAV-NeuroD1-infected areas exhibited a significant decrease (Fig. 7A, B).Microglia release pro-inflammatory cytokines (tumor necrosis factor alpha (TNFα), interleukin 1 beta (IL1β), interleukin 6 (IL6), and inducible nitric oxide synthase (Inos)), which cause neurotoxicity.qRT-PCR analysis showed increased cytokine expression post-HIBI; however, Tnfα, Il1β, and Il6 were significantly downregulated in the AAV-Neu-roD1 group.Inos expression was also weakened, but with no statistical significance, in the AAV-NeuroD1 group (Fig. 7C).
We also investigated the effect of in vivo direct neuronal reprogramming on reactive astrocytes that became active in response to HIBI.Immunostaining revealed decreased GFAP levels in the NeuroD1-treated peri-infarct area at 30 dpi (Fig. 7D, G).Some newly converted neurons (GFP + and NeuN + ) were observed in the peri-infarct area, which is typically devoid of neuronal cells following HIBI (Fig. 7G).qRT-PCR analysis of ipsilateral hemisphere tissue at 30 dpi revealed that the AAV-GFP group had a 12-fold increase in the panreactive astrocyte gene Gfap and an 800-fold increase in the reactive astrocyte-related neuroinflammatory marker lipocalin 2 (Lcn2), whereas the AAV-NeuroD1 group exhibited significant decreases in Gfap and Lcn2 expressions (Fig. 7E).Furthermore, the characteristic upregulation of A1-type reactive astrocyte genes (Serping1, Gbp2, and C3) after HIBI was significantly attenuated by Neu-roD1 treatment (Fig. 7F).By contrast, the expression of the A2-type reactive astrocyte-specific gene S100a10 did not change despite the decrease in GFAP expression after NeuroD1-mediated astrocyte-to-neuron conversion ).E Quantification of the frequency of action potential firing in cortical slices with ischemic injury (blue line, control; red line, NeuroD1 group, n = 5).Data are presented as means ± SEM.Note that the NeuroD1 group showed a significantly higher frequency of action potential firing than the control group (Additional file 4: Fig. S4).These results indicate that NeuroD1-mediated in vivo direct neuronal reprogramming via AAVShH19 attenuated the reactive and neuroinflammatory properties of microglia and astrocytes.

Astrocyte-to-neuron conversion improves function in HIBI
We evaluated whether NeuroD1-mediated astrocyte-toneuron conversion could alleviate HIBI-induced neurobehavioral deficits.Rotarod and adhesive removal tests were conducted at various post-injection time points (Fig. 8).The results of the rotarod test did not differ between AAV-NeuroD1 and AAV-GFP groups at 2 weeks post-injection.However, the AAV-NeuroD1 group spent significantly more time on the rotarod than the vehicle group (248.0 ± 8.9 s vs. 211.4± 15.0 s; Fig. 8A).The performance of the AAV-NeuroD1 group improved over time and was significantly longer than that of the AAV-GFP group at 4 weeks post-injection (275.9 ± 7.1 s vs. 237.2± 14.7 s; Fig. 8A).Rotarod performance did not differ between the AAV-GFP and vehicle groups at any time point (Fig. 8a).The adhesive removal test, which assesses sensorimotor function, was conducted 2-4 weeks postinjection.On week 2, all groups showed similar performance.However, on week 4, the AAV-NeuroD1 group exhibited significantly faster contact (21.4 vs. 39.2 s; Fig. 8B) and removal times (45.0 vs. 75.5 s; Fig. 8C) compared to those of the vehicle group.The AAV-NeuroD1 group also had improved removal times compared with and ####p < 0.0001 vs control.E, F Analysis of relative expression of reactive astrocyte genes Gfap and Lcn2 and A1-type astrocyte-specific genes Gbp2, Serping1, and C3 (n = 3 mice per group).*p < 0.05 vs no injury and #p < 0.05 vs control.G Some newly converted neurons (GFP + and NeuN + cells) were observed in the peri-infarct area in which no neuronal cells are expected to survive after hypoxic-ischemic brain injury.An asterisk marks the infarct area.Dashed lines in g indicate the border of the infarct core.Scale bar, 50 µm.Data are presented as means ± SEM.One-way ANOVA followed by a Bonferroni post hoc test the AAV-GFP group (45.0 vs. 75.5 s; Fig. 8C).Taken together, NeuroD1-mediated astrocyte-to-neuron conversion can alleviate HIBI-induced neurobehavioral deficits.

Discussion
In this study, direct neuronal reprogramming using AAVmediated NeuroD1 gene delivery generated functional new neurons and improved neurobehavioral deficits in vivo in the mouse model of neonatal HIBI.Additionally, in vitro experiments demonstrated that ectopic Neu-roD1 expression in astrocytes could efficiently convert astrocytes into neurons.Functional electrophysiological properties were observed in the newly converted neurons, which integrated with the preexisting host neurons.Furthermore, direct neuronal reprogramming with NeuroD1 protected preexisting neurons from HIBI and reduced toxic A1-type astrocyte gene expression, microglia activity, and neuroinflammation at the injury site.Therefore, in vivo NeuroD1-mediated direct neuronal reprogramming not only induced neuronal regeneration but also protected the microenvironment from further damage in the affected area (Additional file 5: Fig. S5).This overall effect led to a significant improvement in behavioral outcomes.
The direct conversion of glia into neurons has been a prominent research topic in recent years [4,16].Glia cells, abundant in the mammalian brain, have a close lineage to neurons and possess innate proliferative capacity, making them a promising candidate for neuronal reprogramming in the pursuit of neuroregeneration [17].In the past few years, the direct in vivo conversion of reactive astrocytes into mature neurons has proven to be effective in both the adult mouse and non-human primate brain [4,[6][7][8][18][19][20][21].Additionally, strategies exist for direct in vivo conversion of microglia into mature neurons, and it has been shown that microglia in the striatum can be converted into striatal medium spiny neurons [22,23].In our study, we found that the virus infects predominantly specific cortical astrocytes in HIBI, and only a small amount of the virus infects striatal astrocytes, leading to their conversion into striatal medium spiny neurons.However, we confirmed that neurons converted from cortical astrocytes do not express markers of striatal medium spiny neurons, DARPP32 (data not shown).This suggests that the types of neurons generated by in vivo conversion systems may vary depending on the brain region and the characteristics of the target cells.In this study, we examined cortical astrocytes as the target cells for in vivo conversion strategy.Under normal conditions, astrocytes remain in a quiescent state and do not undergo activation or proliferation [24,25], However, in response to brain injuries (e.g., brain ischemia, tumors, mechanical trauma, and neurodegenerative disorders), they become reactive, leading to altered gene expression and accelerated proliferation [14,17,25].Reactive astrocytes are identified by increased GFAP expression [26].To specifically target astrocytes in this study, we employed the GFAP promoter, Cre-FLEX system, and AAV technology.Due to its high infectivity, low pathogenicity, and intrinsic safety, AAV has emerged as a powerful gene delivery tool for treating various human diseases in the field of gene therapy [27,28].The AAV mutant ShH19 exhibited 5.5-fold higher astrocyte transduction than parental AAV2 in the rat striatum; however, only 15% of astrocytes were transduced in a previous study [12].After confirming the distribution of a clear GFP signal in the peri-infarct area through the injection of AAVShH19 encoding GFP into HIBI, the efficiency of specific gene delivery to astrocytes was enhanced using the GFAP promoter (Additional file 6: Fig. S6).Our research data confirmed that hypoxic-ischemic brain injury in the neonatal period triggered the activation of astrocytes around the damaged brain tissue, and the injected NeuroD1 virus predominantly infected and expressed in astrocytes in cortical areas adjacent to the infarct site.The Cre-FLEX system enabled distinguishing between preexisting host neurons and transformed neurons, making it a suitable method for long-term monitoring of transformed neurons that originated from astrocytes [29].
NeuroD1 is a transcription factor that plays a crucial role in embryonic brain development and inducing neuronal differentiation.NeuroD1 induces neuronal fate decisions by directly binding to regulatory elements such as enhancers and promoters of genes involved in neuronal development [30][31][32].Recent works have demonstrated the conversion of reactive astrocytes to glutamatergic neurons through NeuroD1 delivery in a middle cerebral artery occlusion model and Alzheimer's disease model using lentiviruses and retroviruses, respectively [6,33].In a study on spinal cord injury, NeuroD1 overexpression using a retrovirus or AAV resulted in the transformation of astrocytes into neurons [34].Furthermore, NeuroD1 expression through AAV9 reprogrammed astrocytes into both glutamatergic neurons and cortical neurons in a mouse model of stab injury [35].The neuronal conversion of astrocytes was also observed in a rhesus monkey cerebral ischemia model through the delivery of AAV-carrying NeuroD1 [21].Additionally, prior research has demonstrated that signals in vivo, e.g., different immune and inflammatory responses, growth factors, morphogens, and cytokines, can significantly affect different aspects of neuronal conversion [36].Moreover, astrocytes have been observed to differentiate into neurons with different phenotypes via the same transcription factors, depending on the specific brain regions involved [37].It has also been suggested that the intrinsic characteristics of astrocytes in various brain regions, combined with the local environmental niches, collectively play a role in shaping neuronal subtypes during in vivo direct reprogramming [38].
In this study, we conducted the first direct reprogramming investigation on a severe neonatal brain injury, HIBI.We observed that in neonatal mice with HIBI, NeuroD1 predominantly converted astrocytes into GABAergic neurons.This is significant because GABAergic neurons are known to be severely damaged by HIBI [39][40][41], which can decrease the expression of the GABA synthase enzymes GAD65 and GAD67 in the hippocampus after neonatal HIBI [42,43].The dysfunction and loss of GABAergic neurons can impair neural circuits and cognitive function [44].Thus, the recovery of normal GABAergic neuron function in cerebral ischemia has been identified as a potential strategy to protect neurons and improve cognitive outcomes [44].As the exact mechanism underlying the production of GABAergic neurons has not been fully elucidated in this study, further investigation through single-cell chromatic analysis is needed in further research.
The converted neurons in the NeuroD1 group exhibited electrophysiological properties and expressed synaptophysin, suggesting that they established synaptic connections.These results suggest that NeuroD1-mediated conversion from astrocytes to functional neurons in vivo could replace lost neurons and improve brain function affected by HIBI.
In our study, NeuroD1 consistently converted astrocytes into GABAergic neurons both in healthy astrocytes (in vitro) and astrocytes within the peri-infarct area of hypoxic-ischemic brain injury (in vivo).These results contrast with previous studies using adult disease models of ischemic stroke or other brain injury, where cortical astrocytes were dominantly converted into glutamatergic neurons by NeuroD1 [45].We hypothesize that delivery of NeuroD1 to cortical astrocytes during developmental stage directs their fate towards GABAergic neurons, while there are still limitations in elucidating the exact mechanism.Indeed, the density of GABAergic neurons in the cerebral cortex and white matter rapidly increases over the second half of gestation, peaking at term and then declining during the first six months of life, suggesting that the perinatal period is a critical time for the development of GABAergic system [46,47].Additionally, reprogramming with the vector AAVShH19 with the human GFAP promoter was attempted for the first time in our study, so the fate of reprogrammed cells may be differently triggered compared to previous research.Recently, Xi et al. reported that when NeuroD1 was delivered to Müller cells using two different serotypes of AAV (AAV7m8 and AAV9, with different GFAP promoters), the final converted cells were also different [48].To figure out the exact mechanism underlying the production of GABAergic neurons has not been fully elucidated in this study, further investigation through single-cell chromatic analysis is needed in further research.
Brain disorders cause activation of astrocytes, i.e., reactive astrocytes, which subsequently undergo reactive astrogliosis and form a glial scar around the injury site.Reactive astrocytes also secrete growth inhibitory and inflammatory factors, e.g., TNFα, Lcn2, and proteoglycan proteins, which impede neural repair and regeneration.As brain injuries progress, the glial scar becomes a significant physical and chemical barrier that hinders tissue repair and neuronal regeneration.This barrier can persist for a long time and greatly impairs brain function recovery, particularly during the later stages of brain injuries [15,[49][50][51].Recent studies have categorized reactive astrocytes based on their functions and gene expression patterns into A1-type and A2-type astrocytes.A1-type astrocytes have detrimental effects on central nervous system damage, such as neuronal cell death and neurotoxic cytokine secretion, whereas A2-type astrocytes secrete neuroprotective factors and promote neuronal survival [52,53].Our data shows that the number of reactive astrocytes and the expression levels of A1-type astrocyte-specific genes (Serping1, Gbp2, and C3) [54,55] significantly increased after HIBI but were significantly decreased in the NeuroD1-treated group.This indicates that direct NeuroD1-mediated neuronal reprogramming can help suppress the detrimental effects of reactive astrocytes in HIBI.In addition to astrocytes, microglia also accumulate rapidly around the lesion in HIBI and increase the expression of pro-inflammatory factors and neurotoxic cytokines that inhibit axonal regeneration.This activation of microglia contributes to secondary damage to neurons and inhibits the recovery process [56,57].Our study found that the expression levels of proinflammatory cytokines and the activation of microglia in the damaged cortex were significantly reduced in the AAV-NeuroD1-injected group compared with the AAV-GFP-injected group (Fig. 7A-C).Although it was not clear which cellular response occurred first, it is possible that the decreased number of reactive astrocytes and decreased microglia activation are related.
Previous studies have shown that neurotoxic reactive astrocytes can be induced by cytokines (e.g., IL1a, TNFα, and component 1 subcomponent q) that are released by reactive microglia [53,58].Astrocytes can also regulate microglial activity through various factors, such as Ca 2+ , ATP, chemokines, complement (C3), and plasma protein orosomucoid 2 [59][60][61][62][63].By promoting direct astrocyteto-neuron conversion, NeuroD1 indirectly modulates the reactivity of microglia and astrocytes at the site, leading to a reduction in neuroinflammation.This reduction in neuroinflammation improves the histological and functional outcomes of brain disease [54,[64][65][66].Therefore, the protection of preexisting host neurons around the damaged area after NeuroD1-mediated neuron conversion observed in this study may be affected by microenvironmental changes, such as decreased numbers of reactive astrocytes and microglia, downregulated A1-type astrocyte genes, and decreased levels of pro-inflammatory factors.Direct Neu-roD1-mediated neuronal reprogramming in vivo resulted in these effects, which demonstrated improvements in neurobehavioral function in an animal model of neonatal HIBI.These findings suggest that direct neuronal reprogramming may have therapeutic potential for treating brain injuries and diseases by promoting neural repair and regeneration while also modulating the immune response to injury.However, additional studies are required to fully understand the underlying mechanisms that produce these effects and to determine the clinical applicability of direct neuronal reprogramming in the treatment of brain diseases.

Conclusion
To summarize, NeuroD1-mediated in vivo neuronal conversion via AAVShH19 efficiently converted astrocytes to neurons with functional electrophysiological properties, providing a potential strategy for replacing lost neurons caused by neonatal HIBI.Moreover, this approach reduced neuroinflammation and protected preexisting neurons, thereby mitigating sensorimotor deficits.Our findings suggest that in vivo neural regeneration via AAV-NeuroD1 gene delivery to astrocytes may represent a therapeutic strategy for neuronal regeneration in cases of neonatal HIBI.

Fig. 5
Fig. 5 NeuroD1 converted astrocytes into GABAergic neurons in the hypoxic-ischemic-injured brain.A Expression of the GABAergic neuronal marker GAD67 or GABA (red) in GFP + cells (green; overlay color: yellow/orange) 4 weeks after NeuroD1 transduction.Scale bar, 25 µm.B Expression of the GABAergic neuronal markers GAD67 or GABA (red) in converted neurons showing co-expression of GFP + (green) and NeuN + (blue) 6 weeks after NeuroD1 transduction.Scale bar, 25 µm.C, D A high-magnification confocal image and the three-dimensional reconstruction of the image showing a converted neuron co-stained with GFP, NeuN, and GAD67.Arrowheads indicate co-labeled cells.Scale bar, 50 µm.E Quantitative analysis of GFP + cells co-labeled with GAD67 or GABA.Note that 55.7% and 21.7% of converted cells express GAD67 and GABA, respectively (n = 3 mice per group).Data are presented as means ± SEM. **p < 0.01, ***p < 0.001 vs control

Fig. 6
Fig. 6 Neurons converted by NeuroD1 were physiologically functional.A, B Synaptic marker (synaptophysin, red) expression on the newly converted neurons (GFP + and NeuN + , overlay color: light blue) in AAV-NeuroD1-infected cells (green, 60 dpi).A high-magnification confocal image and the three-dimensional reconstruction of the image showing a converted neuron co-stained with GFP, NeuN, and synaptophysin.Arrowheads indicate co-labeled cells.Scale bar, 50 µm.C Phase and fluorescent images of a converted neuron.Arrowheads indicated converted cells.Scale bar, 20 µm.D NeuroD1-converted neurons displayed repetitive action potential firing (60 dpi, n = 12).E Quantification of the frequency of action potential firing in cortical slices with ischemic injury (blue line, control; red line, NeuroD1 group, n = 5).Data are presented as means ± SEM.Note that the NeuroD1 group showed a significantly higher frequency of action potential firing than the control group

Fig. 7
Fig. 7 Astrocyte-to-neuron conversion by NeuroD1 attenuated microglial inflammatory responses and reduced glial scar after hypoxic-ischemic brain injury.A Low-magnification (left) and high-magnification (right) fluorescence images showing a decrease in microglial (Iba1, red) number in the ipsilateral hemisphere of the AAV-NeuroD1 group (NeuroD1, bottom) compared with the control AAV-GFP group (control, middle).Scale bar, 400 µm (left panels, low magnification); 100 µm (right panels, high magnification).B Quantitative analysis of the number of microglia around the injury sites (n = 3-5 mice per group).***p < 0.001 vs no injury and ###p < 0.001 vs control.C Analysis of pro-inflammatory cytokine mRNA levels in hypoxic-ischemic-injured ipsilateral hemisphere at 4 weeks post-injection (n = 3-5 mice per group).*p < 0.05, **p < 0.01, ***p < 0.001, *****p < 0.00001 vs no injury and #p < 0.05, ###p < 0.001, ####p < 0.0001 vs control.Images in G showing accumulated astrocytes expressing GFAP (red).D Cell numbers were significantly decreased in the AAV-NeuroD1 group (n = 3 mice per group).*p < 0.05, *****p < 0.00001 vs no injury and ####p < 0.0001 vs control.E, F Analysis of relative expression of reactive astrocyte genes Gfap and Lcn2 and A1-type astrocyte-specific genes Gbp2, Serping1, and C3 (n = 3 mice per group).*p < 0.05 vs no injury and #p < 0.05 vs control.G Some newly converted neurons (GFP + and NeuN + cells) were observed in the peri-infarct area in which no neuronal cells are expected to survive after hypoxic-ischemic brain injury.An asterisk marks the infarct area.Dashed lines in g indicate the border of the infarct core.Scale bar, 50 µm.Data are presented as means ± SEM.One-way ANOVA followed by a Bonferroni post hoc test

Fig. 8
Fig. 8 Neuronal conversion mediated by NeuroD1 improved functional outcomes after hypoxic-ischemic brain injury.A-C After injection of AAV-NeuroD1 (NeuroD1, n = 21); control AAV-GFP (control, n = 19); and vehicle (n = 18), rotarod (A) and adhesive tape removal (B, C) tests were performed for behavioral analysis.The rotarod test was performed 2, 3, and 4 weeks post-injection, and the adhesive tape removal test was performed 2 and 4 weeks post-injection.Data are presented as means ± SEM. *p < 0.05, **p < 0.001 vs vehicle and #p < 0.05 vs control.One-way ANOVA followed by a Bonferroni post hoc test