Single-cell Transcriptomic Profiling of Brains in Newborn Rats Following Hypoxic Ischemic Encephalopathy

doi:10.21203/rs.3.rs-4988385/v1

Download PDF

Research Article

Single-cell Transcriptomic Profiling of Brains in Newborn Rats Following Hypoxic Ischemic Encephalopathy

https://doi.org/10.21203/rs.3.rs-4988385/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Neonatal hypoxic-ischemic encephalopathy (HIE) is a severe neurological condition associated with high rates of mortality or long-term disability. Despite its clinical significance, the detailed cellular mechanisms underlying HIE remain unclear. Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful tool for investigating cellular heterogeneity across development, aging, and disease processes. However, no scRNA-seq studies have yet addressed neonatal HIE.

Methods

We employed scRNA-seq to examine cellular heterogeneity during the hyperacute (3 hours), acute (2 days), and subacute (7 days) phases of neonatal HIE. Uniform Manifold Approximation and Projection (UMAP) was used to visualize the cell clustering. Differentially expressed genes (DEGs) were calculated and identified using the Seurat’s FindAllMarkers function, which was enriched for pathway analysis (GO, KEGG pathway, WikiPathways, and Reactome Gene Sets). CytoTRACE v2 was used to identify the maturity state of each cell type and pseudotime analysis was performed using Monocle v3.

Results

We analyzed a total of 87,580 high-quality brain cells to identify transcriptional changes associated with HIE. In the hyperacute phase, we observed activation of astrocytes in response to reactive oxygen species, involvement of microglia in phagocytosis, Stat3-mediated ischemic responses in oligodendrocyte precursor cells, and an increase in senescent lymphatic endothelial cells. In the acute phase, astrocytes were found to exacerbate inflammation and impede brain development, while microglia proliferated. Neuroblasts were affected by metal ions, and oligodendrocytes decreased. In the subacute phase, astrocytes facilitated tissue repair, while inflammatory microglia highly expressing MHC II were induced by the IL27 and type I interferon pathways and expanded. Additionally, peripheral immune cells played vital roles in HIE. Specifically, neutrophils infiltrated and expanded throughout all phases post-HIE. Spp1^high macrophages, T cells, and plasmacytoid dendritic cells increased during the acute and subacute phases, and B cells expanded during the subacute phase.

Conclusion

This study offers deep insights into the molecular alterations of key cell types following HIE, elucidating the pathological processes involved. These findings have significant implications for developing effective clinical strategies for managing HIE.

Hypoxic-ischemic encephalopathy

HIE

Single cell RNA sequencing

ScRNA-seq

Newborn rats

Neonatal hypoxic-ischemic encephalopathy (HIE) is a severe neurological injury occurring after birth, leading to high rates of mortality and long-term neurological disability¹. It affects approximately 2–6 per 1000 newborns in developed countries and 26 per 1000 in developing countries, with significant mortality and disability rates^2–4. Since the occurrence and development of HIE is a continuous process, the pathology of HIE progresses through three main phases²: i) The primary injury (0–6 h), characterized by rapid ATP depletion due to ischemia and hypoxia, leading to decreased activity of ATP-dependent ion pumps, Ca²⁺ accumulation in neurons, and neuronal edema⁴. Excitotoxicity and lactic acid accumulation further contribute to neurotoxicity^5,6; ii) The secondary injury (6–72 h), marked by neuronal death shortly after reperfusion. Initial excitotoxicity exacerbates Ca²⁺ influx, triggering a cascade of events including lipase activation, mitochondrial dysfunction, nNOS activation, fatty acid release, and free radical production². Mitochondrial dysfunction is central, leading to ROS accumulation, which activates microglia and astrocytes and promotes inflammation⁷. In addition, excessive ROS induces mitochondrial permeability transition pores (mPTP), facilitating the release of pro-apoptotic factors and cytochrome c, ultimately driving apoptosis⁸; iii) The persistent injury (72 h - several weeks to years), where ongoing inflammation is predominant. This phase involves resident cells like microglia and astrocytes, as well as infiltrating immune cells such as neutrophils, monocytes, macrophages, and lymphocytes^9–11. Persistent inflammation leads to continued neuronal death and epigenetic changes that impair oligodendrocyte maturation, axonal growth, and neurogenesis¹².

Currently, hypothermia, which lowers body and brain temperatures to approximately 32–34°C, is the only standard therapy for neonatal HIE. This approach offers neuroprotection by reducing ROS, free radicals, and excitatory amino acids, inhibiting inflammation, and preventing neuronal apoptosis^13–15. Hypothermia at 33.5°C for 72 hours is optimal for neonates with moderate to severe HIE, though its neuroprotective effects are limited^16–18. Therefore, exploring additional therapies, such as drugs, cell therapy, and gene therapy, is urgent¹⁹. However, comprehensive understanding of the cellular mechanisms underlying HIE remains elusive, hindering the development of effective clinical targets.

Single-cell RNA sequencing (scRNA-seq) is emerging as a powerful technology for exploring cellular heterogeneity across development, aging, and disease processes, facilitating insights into cellular dynamics and identifying novel cell subpopulations^20–22. ScRNA-seq has been applied to various central nervous system injuries and diseases, including spinal cord injury, brain injury, stroke, Alzheimer’s disease, and multiple sclerosis, uncovering novel injury/disease-associated cell types^20,23–26. However, no scRNA-seq studies have yet addressed neonatal HIE. Given that neonatal brain cells are more naïve compared to those in adult ischemic stroke, exploring cellular heterogeneity in neonatal HIE through scRNA-seq is warranted.

In this study, we performed scRNA-seq to characterize cell type composition and transcriptional dynamics during the hyperacute (3 hours, primary injury), acute (2 days, secondary injury), and subacute (7 days, persistent injury) phases of neonatal HIE. Our findings identified novel cell subpopulations and advanced the understanding of cellular mechanisms involved in HIE.

Animals

All animal experiments were performed in accordance with the China National guidelines for the care and use of laboratory animals and were approved by the Ethics Committee of Zhejiang University (No. ZJU20230229). Sprague–Dawley (SD) rats were housed at the Laboratory Animal Center of Zhejiang University with ad libitum access to sterile water and food.

Models of HIE

Models of HIE were established in postnatal day 7 (P7) SD rat pups, with no gender preference, following a previously described protocol with minor modifications²⁷. Briefly, rat pups were randomly assigned to six groups (three males and three females per group): Sham-3h, HIE-3h, Sham-2d, HIE-2d, Sham-7d, and HIE-7d. The pups were deeply anesthetized using isoflurane inhalation. The right common carotid artery was surgically exposed and doubly ligated in the HIE groups. Sham-operated rats underwent identical exposure of the right common carotid artery without ligation. The incision was sutured, and the pups were returned to their dams. Subsequently, the HIE rat pups were placed in a chamber with 8% oxygen for 1.5 hours at 37°C.

Single cell dissociation

After 3h, 2d, and 7d, Sham and HIE rat pups were anaesthetized as above describe and perfused with ice-cold saline transcardially. Ischemic and hypoxic brains were quickly dissected and maintained in artificial cerebrospinal fluid (aCSF). Brains were cut into around 1mm x 1 mm small pieces and digested in aCSF containing 1 mg/mL papain (LS003120, Washington, USA), 1 mg/mL collagen II (LS004176, Washington, USA), and 50 IU/mL DNase (EN0251, ThermoFisher Scientific, USA) for 30 min at 37℃. After digestion, single cell suspension was generated by gentle pipette blowing and filtered through a 70 µm cell strainer. After centrifugation, cells were resuspended in 33% Percoll (17089109, Cytiva, Sweden) and centrifuged to remove myelin debris.

Single cell RNA sequencing

Isolated single cells were resuspended in PBS buffer at a concentration of 3× 10⁵ cells/mL and next loaded onto a microfluidic chip (GEXSCOPE® Single Nucleus RNA-Seq Kit, Singleron, China). The preparation of scRNA-seq libraries was conducted according to the manufacturer’s protocol and the library quantification was conducted using Qubit. Sequencing was performed on an Illumina NovaSeq 6000 instrument with 150 base pair paired-end reads and gene-barcode count matrices were generated using the recommended pipeline of CeleScope. The matrices were further processed using Seurat v4.2²⁸. Cells with more than 5% mtDNA or fewer than 400 distinct genes were removed in our scRNA-seq data. Then, sequencing data was scaled and normalized using a Seurat’s LogNormalize method. Highly variable genes in each cell were recognized using the Seurat’s FindVariableGenes function. Cell cycle status were identified using a Seurat’s CellCycleScoring method and the influence of cell cycle on clustering was eliminated. Multiple scRNA-seq matrices were integrated and the batch correction was conducted using Harmony²⁹. The data was next performed using a principal component analysis (PCA) method and clustering was performed using a FindClusters method. Uniform Manifold Approximation and Projection (UMAP) was used to visualize the cell clustering. Expressed genes were calculated and identified using the Seurat’s FindAllMarkers function, and the differentially expressed genes (DEGs) were defined as genes with fold change > 1.5 and adjust p value < 0.05. Pathway enrichment analysis (GO, KEGG pathway, WikiPathways, and Reactome Gene Sets) of DEGs was conducted using Metascape (http://metascape.org)³⁰. CytoTRACE v2 was used to identify the maturity state of each cell type and pseudotime analysis was performed using Monocle v3.

Single-cell profiling of the brain after HIE

To investigate the cellular heterogeneity associated with HIE, scRNA-seq was performed on brain samples isolated from HIE rat pups at three-time points (3 hours, 2 days, and 7 days) and their corresponding sham controls using the Singleron platform. After rigorous quality filtering, a total of 87,580 cells were analyzed, including 12,995 cells from the sham-3h group, 16,650 from the HIE-3h group, 16,364 cells from the sham-2d group, 13,722 cells from the HIE-2d group, 14,801 cells from the sham-7d group, and 13,048 cells from the HIE-7d group (Fig. 1A). Fourteen distinct clusters were identified and visualized using Uniform Manifold Approximation and Projection (UMAP) graph, encompassing astrocytes (25,189 cells), microglia (19,310 cells), neuroblasts (11,072 cells), endotheliocytes (8,028 cells), oligodendrocytes (OCs, 7,475 cells), oligodendrocyte precursor cells (OPCs, 6,659 cells), mononuclear phagocytes (MPs, 2,664 cells), mural cells (2,378 cells), fibroblast (2,314 cells), ependymal cells (1,085 cells), choroid plexus cells (CPCs, 603 cells), red blood cells (RBCs, 340 cells), neuroendocrine cells (NCs, 267 cells), and neurons (196 cells) (Fig. 1A).

Cell type identification was dependent on established signature markers as previously described^20,31. Specifically, astrocytes were characterized by high expression of Aqp4, Gja1, Gfap, and Plpp3. Microglia expressed Hexb, P2ry12, Tmem119, and Cx3cr1. Neuroblast were marked by Tubb3, Dcx, Stmn2, and Dlx1. Endotheliocytes exhibited Cldn5, Flt1, Itm2a, and Vwf expression. OCs were identified by Plp1, Mbp, Mag, and Mog. OPCs expressed Pdgfra, Olig1, Olig2, and C1ql1. MPs were distinguished by macrophage-specific marker genes (Pf4, Mrc1, and Cd68) and a monocyte-specific marker gene (Lyz2). Mural cells expressed pericyte-specific marker genes (Abcc9, Pdgfrb, Rgs5, and Kcnj8) and vascular smooth muscle cell-specific marker genes (Mylk and Myl9). Fibroblast were marked by Dcn, Col3a1, and Col1a1. Ependymal cells expressed Dynlrb2, Ccdc153, Ak9, and Ak7. CPCs were identified by Folr1, Kcnj13, and Clic6. RBCs exhibited Hbb, Hbb-a2, and Hbb-a3 expression. NCs expressed Chga, Gnb3, and Gnat2. Neurons were characterized by Slc17a6, Nhlh2, and Map2 (Fig. 1B and Table S1).

Doughnut diagrams illustrated the proportions of these clusters in each group (Figs. 1C and S1). At the P7 stage, the sham-3h group exhibited a higher proportion of naive cells, such as OPCs and neuroblasts. By P9 (sham-2d), the proportion of astrocytes increased, while OCs increased by P14 (sham-7d). During the hyperacute phase of HIE (3 h), proportions of astrocytes and OPCs decreased, whereas MPs increased. During the acute phase (2 d), the proportions of astrocytes, neuroblasts, and OCs decreased, while microglia and MPs increased. During the subacute phase (7 d), the proportions of astrocytes, endotheliocytes, and OCs decreased, while microglia, neuroblast, and MP increased.

Dynamic changes of astrocytes following HIE

To understand the dynamic changes in astrocytes during HIE, we performed re-clustering of astrocytes, identifying 12 distinct subtypes (A1-A12) (Fig. 2A). DEGs analysis showed that A1 upregulated Slco1c1, Dbndd2, Scg3, Rgcc, Ca8, et al; A2 upregulated Adgrb3, Plp1, Gabrb1, Nrxn1, et al; A3 upregulated Agt, Timp4, Kcnj16, Igsf1,Slc6a9, et al; A4 upregulated Tspan7, Nr2f2, Hsd11b1, Olfm3, Lhx2, et al; A5 upregulated Serpinf1, Id3, Lcat, Emb, Olfm1, et al; A6 upregulated Gfap, Vim, Igfbp5, Tagln, Ifitm3, et al; A7 and A9 upregulated Prc1, Tpx2, Cdk1, Top2a, Cdkn3, et al; A8 upregulated Igfbp4, Egr1, Fos, Ier2, Junb, et al; A10 up-regulated Sfrp1, Penk, Veph1, Sfrp2, Hgf et al; A11 upregulated Ier3, Tyrobp, Ltc4s, C1qb, Ccl3, et al; A12 upregulated Epha3, Cldn9, Draxin, Ccdc80, Fgfbp3 et al (Fig. 2B and Table S2). Pathway enrichment analysis of DEGs (Figure S2A-D and F-I) and cell cycle analysis (Figure S2E) allowed us to categorize A1 as inhibitory astrocytes, A2 as morphogenetic astrocytes, A3 as angiogenesis astrocytes, A4 as Nr2f2^high astrocytes, A5 as metal ion reactive astrocytes-A type, A6 as metal ion reactive astrocytes-B type, A7 as S phase astrocytes, A8 as translation active astrocyte, A9 as G2M phase astrocytes, A10 as metal ion reactive astrocytes-C type, A11 as inflammatory astrocytes, and A12 as repair-related astrocytes (Fig. 2A).

The UMAP distribution of each sample across cell cluster categories is shown in Fig. S2J. CytoTRACE analysis ranked cell maturity in the order of clusters A9, A11, A8, A7, A6, A10, A12, A5, A4, A2, A1, and A3 (Figure S2K). Based on this analysis, we designated A9 as the starting point for the astrocytes' single-cell trajectory path (Fig. 2C). A heatmap depicting co-regulated gene modules in each astrocyte sub-cluster is shown in Fig. 2D and Table S3. Notably, A1 was related to A4, A2 was related to A11, A7 was related to A9, and A6 was related to A10. A3 upregulated modules 55 and 56, A5 upregulated module 34, A2 upregulated modules 54 and 35, A7 upregulated modules 10 and 39, A9 upregulated module 9, A12 upregulated module 65, A6 upregulated module 13, and A10 upregulated modules 32 and 21.

Figure 2E illustrates the proportion of each group among all astrocyte subtypes. Notably, A8 and A10 were elevated in mice at 3 hours post-HIE compared to sham mice (29.5% vs 21.3% and 30.76 vs 22.6%, respectively). As mentioned above, besides translation-associated genes, A8 up-regulated Igfbp4, and immediate early genes (IEGs) including Egr1, Fos, Junb, Jun, and Jund. In addition, A10 showed increased expression of ROS detoxification-related genes (Gstp1, Sod1, Cyba, and Prdx1) and Wnt signaling pathway-associated genes (Sfrp1, Sfrp2, and Ankrd6). A6, A7, and A11 were more prevalent in mice at 2 days post-HIE compared to sham controls (42.47% vs 11.23%, 38.29% vs 16.07%, and 19.29% vs 6.26%, respectively). We noticed that metal ion reactive astrocytes-B type (A6) also upregulated Igfbp5. Furthermore, A3, A11, and A12 increased in mice at 7 days post-HIE compared to their sham mice (36.35% vs 20.49%, 39.22% vs 12.9%, and 26.88% vs 12.43%, respectively). A3 was enriched for angiogenesis associated genes (such as Hif1a, Ramp1, Angptl4, Tmem100, and Nrarp), as well as Agt, Timp4, and Kcnj16.

In conclusion, astrocytes rapidly activated in response to ROS during the hyperacute phase of HIE. During the acute and subacute phases, astrocytes inhibit brain development by suppressing the IGF pathway, showing an inflammation-related phenotype in the acute phase and a repair-associated phenotype in the subacute phase.

Dynamic changes of microglia following HIE

Microglia were subsequently collected and re-clustered into 10 subtypes (M1-M10) (Fig. 3A). According to DEGs (Fig. 3B and Table S4), M1 expressed high levels of homeostatic microglia marker genes such as Tmem119, Sparc, P2ry12, Siglec5, Hexb, and Cx3cr1, classifying it as homeostatic microglia^20,32. M2 upregulated phagocytic genes such as Cd68 and lysosome related genes such as Cd63, Ctsb, Atp6v0c, Ctsd, Dnase2, Ctsz, Npc2, Cd68, Ctsa, and Hexa, defining it as phagocytic microglia³³. Cell cycle analysis identified M3 and M6 as G2M phase microglia, with M4 as S phase microglia (Figure S3A). Notably, M6, characterized by oxidative stress-related genes including Gpx1, Gpx3, Banf1, Txn1, and Ppial4d, was classified as stress-related G2M phase microglia. Some microglial subtypes (M5, M8-10) co-expressed other cell type marker genes. M5 were enriched with astrocytic genes (Aqp4, Gja1, S100b, Gfap, Sox9 et.al), M8 with oligodendroglia genes (Mog, Mag, Mobp, Mbp, Plp1, et.al), M9 with OPC specific genes (Scrg1, Olig1, Olig2, Sox10, Pdgfra, et.al), M10 with neuronal genes (Tubb3, Stmn2, Dcx, Map2, Map1b, et.al). In addition, M7 upregulated MHC II- related genes (Cd74, RT1-Db1, RT1-Da, RT1-Ba, RT1-Bb, et. al) and inflammatory genes (C3, Cxcl16, Ccl6, Cxcl2, Ccrl2, Ccl4, et.al), which was defined as MHC II^high inflammatory microglia.

The UMAP distribution of each sample for each microglial subcluster category is shown in Fig. S3B. The quantity of microglia increased during development, remaining stable during the hyperacute phase of HIE but rising during the acute and subacute phases. Figure 3C depicts the proportions of each microglial subtype over time. Homeostatic microglia (M1) increased with development (Sham-3h (P7): 10.99%, Sham-2d (P9): 19.16%, Sham-7d (P14): 29.77%), but decreased during HIE (HIE-2d: 11.99%, HIE-7d: 17.51%). Phagocytic microglia (M2) decreased with development (Sham-3h (P7): 17.79%, Sham-2d (P9): 11.26%, Sham-7d (P14): 8.99%), but increased after HIE (HIE-3h: 23.77%, HIE-2d: 11.99%, HIE-7d: 17.51%). Notably, neuronal gene-enriched microglia (M10) increased following HIE at 3 h compared to their corresponding Sham group (29.12% vs 19.26%). The proportion of G2M phase and stress related G2M phase microglia increased in the HIE-2d group compared to their corresponding Sham group (27.73% vs 15.51% and 26.53% vs 18.7%). MHC II^high inflammatory microglia (M7) reached up to 73.44% at 7 days post-HIE.

CytoTRACE analysis predicted M3 as the starting point of the microglial single-cell trajectory path (Figure S3C). Monocle (v3) analysis suggested that M7 may originate from M2 (Fig. 3D). Inflammatory microglia (M7) were prominent in the HIE-7d group, potentially crucial for HIE progression. DEG interactions in M7 were analyzed by STRING, with the top 10 associated genes being Stat1, Cd74, Irf7, Isg15, Ifi44, Usp18, B2m, Cd68, Rtp4, and Ifit2 (Figure. 3E). STRING analysis revealed that these inflammatory microglia were linked to the IL27 and IFN I type pathways.

Dynamic changes of neuroblast and oligodendrocyte lineage following HIE

Re-clustering of neuroblasts was performed to elucidate the dynamic changes in neuroblasts following HIE. We successfully obtained a total of 12 subclusters of neuroblasts (NB1-12) (Fig. 4A). Specifically, DEGs and cell cycle analysis (Figures S4A and 4B, and Table S5) revealed that NB1 and NB12 exhibited high expression of genes associated with GABAergic synapse, such as Grad1, Grad2, Gnao1, Slc32a1, and Gabarapl, and were thus categorized as GABAergic neuroblasts. NB2 was recognized as S-phase neuroblasts. NB3, which upregulated genes related to metal ion homeostasis, was classified as metal ion-responsible neuroblasts. NB4 and NB6, which upregulated genes associated with AMPA receptors, including Gria2, Nrn1, Olfm1, and Gnih2, were identified as AMPAR neuroblasts. NB5 and NB11 were defined as G2M-phase neuroblasts. NB7 significantly expressed lipid metabolism associated genes, which was defined as lipid metabolism related neuroblasts. NB8 was recognized as G2M/S-phase neuroblasts. NB9, which upregulated genes related to glutamatergic synapses, including App, Gira2, and Atp1a3, was classified as glutamatergic neuroblasts. NB10 upregulated dopaminergic synapse related genes including Plcb1, Ppp1ca, Prkcb, Kcnj6, Cria4, Gng3, and Kif5c, which was defined as dopaminergic neuroblasts. The proportion of metal ion-responsible neuroblasts (NB3) increased during the acute and subacute phases of HIE (24.79% vs 8.69%, 39.34% vs 6.69%) (Figs. 4C and S4B).

To further investigate the dynamic changes in oligodendrocyte lineage cells following HIE, OPCs and OCs were re-clustered into 12 subclusters, including five subtypes of OCs, six subtypes of OPCs, and one intermediate OC/OPC subtype (Figs. 4A and S4C). Based on significant DEGs identified in each subcluster and subsequent enrichment pathway analyses (Figs. 4E, S4D and E, and S5A, Tables S7-9), the subclusters were characterized as follows: OC1 as antigen presentation-related OCs, OC2 as scavenger receptor-active OCs, OC3 as ER stress-related OCs, OC4 as growth-related OCs, OC5 as lipid metabolism-related OCs, OPC/OC as intermediate state OPC-OC cells, OPC1 as cell adhesion-type OPCs, OPC2 as NMD-related OPCs, OPC3 as S-phase OPCs, OPC4 as G2M-phase OPCs, OPC5 as ischemia-responsible OPCs, and OPC6 as neurogenesis-related OPCs (Fig. 4D). It was observed that OPCs progressively transitioned into OCs during development (Figure S5B). During the acute and subacute phases of HIE, the number of OCs decreased while the number of OPCs remained unchanged, suggesting that HIE may lead to myelin damage or impairment of myelin maturation. Notably, a significant enrichment of ischemia-responsible OPCs was observed following HIE, particularly during the hyperacute phase (HIE3h-67.05%; HIE2d-15.57%; HIE7d-11.98%) (Fig. 4F). Ischemia-responsible OPCs upregulated Vgf (Fig. 4G). Interaction network analysis of DEGs suggested that Stat3 might be a critical transcription factor for ischemia-responsible OPCs (Figure S6). In addition, the proportions of Pdcd4^high OC1 and Cdkn1c^high OC3 slightly increased during the acute phase of HIE (22.75% vs 17.17%, 14.55% vs 9.67%, respectively) (Fig. 4G).

Dynamic changes of endotheliocytes and fibroblasts following HIE

Re-clustering of endotheliocytes was conducted to explore dynamic changes of these cells following HIE. Thirteen subpopulations were identified, including six subclusters of vascular endothelial cell (VEC), five subclusters of lymphatic endothelial cell (LEC), and one subcluster of mixed cells were identified (Fig. 5A). According to a previous study and DEGs (Figs. 5B and S7A, Table S10)³⁴, we identified Ca4^high VEC-capillary that also highly expresses Nrgn, Tfrc, Slc22a8, and Hmcn1, and Rgcc^high VEC-capillary that highly expresses Aqp4, Ptprz1, Slc1a3, Mt3, and Fabp7. In addition, we identified Sox17^high Gja4^high VEC-arterial, Mdcam1^high VEC-venous, VEC-inter that co-expresses VEC-arterial and VEC-venous marker genes, and a novel VEC subpopulation that highly expresses Dcn, Slc7a11, Igf2, Col1a1, and Ptgds. For LEC, we identified Lyve1^low LEC-major, Lyve1^high LEC-major, Scg3^high LEC-valve, Cldn11^high LEC-valve, and Cd24^high LEC-valve. We noticed that the proportion of LEC-valve3 increases during the hyperacute phase of HIE (41.49% vs 15.01%) (Figs. 5A and S7B). Enrichment pathway analysis of significant DEGs in LEC-valve3 showed that these LECs were independently related to cellular senescence with the high expression of Jun, Cdkn1b, Cdk4, Ube2s, Ube2c, Ezh2, Fos, H2ax, and H2az2.

Fibroblasts were also re-clustered into nine subtypes (FIB1-9) (Fig. 5D). FIB1 was defined as migrated fibroblasts, expressing Sparcl1; FIB2, as adhesion-related fibroblasts, expressing Wnt6; FIB3 as tissue remodeling-related fibroblasts, expressing Tnnt2; FIB4 and FIB9, as proliferative fibroblasts, with FIB4 expressing Depdc1 and FIB9 expressing Igfbpl1; FIB5, as energy metabolism-activated fibroblasts, expressing Tnnt2; FIB6 expressing Sox10; FIB7, as fibroblasts related to carboxylic acid transport, expressing Tmem229a; and FIB8, as activated fibroblasts, expressing Gpr34 (Fig. 5E and Table S11). Changes in fibroblast proportions were observed primarily during the subacute phase of HIE, with increases in FIB2-5 (22.47% vs 13.79, 34.16% vs 19.36%, 26.01% vs 12.52%, 17.37% vs 11.29%, respectively) (Figs. 5F and S7D).

Dynamic changes of infiltrated immune cells following HIE

Re-clustering of the MP compartment revealed the presence of various immune cell types, including neutrophils, mast cells, B cells, T cells, and dendritic cells (DCs), in addition to macrophages and monocles (Fig. 6A). These cells were classified as infiltrated peripheral immune cells. Macrophages specifically expressed markers such as Mrc1, Pf4, and Aif1 (Fig. 6B). Other cell type marker genes enriched (OCTMGE) macrophages upregulated astrocyte, oligodendrocyte, and neuron maker genes such as Sox9, Olig1, and Tubb3. MHCII^high macrophages upregulated Cd74. Monocytes specifically expressed Treml4. T cells specifically expressed Cd3e and Cd2. Activated T cells upregulated activated marker genes including Cd69 and Icos. Neutrophils specifically expressed S100a9 and Mmp9. Mast cells specifically expressed Fcer1a. B cells specifically expressed Cd19. Plasmacytoid dendritic cells (pDCs) specifically expressed Ccr9.

We identified four macrophage subtypes, homeostatic macrophages highly expressed macrophage marker genes including Aif1, Pf4, and Mrc1, whereas OCTMGE macrophages, Spp1^high macrophages, and MHCII^high macrophages down-regulated these marker genes. The proportion of OCTMGE macrophages increased 2 days after HIE (24.64% vs 15.25%), which may indicate that these macrophages phagocytose other cells containing these mRNA (Fig. 6C). Importantly, Spp1^high macrophages increased during acute and subacute phases of HIE (Acute, 42.43% vs 5.9%; Subacute, 29.13% vs 13.25%) (Fig. 6C), which was related to lysosome, phagosome, TNF production, detoxification of reactive oxygen, lipid metabolism, response to stress (Fig. 6D). MHCII^high macrophages increased following HIE (Hyperacute, 9.09% vs 2.33%; Acute, 33.09% vs 7.4%; Subacute, 34.8% vs 13.29%) (Fig. 6C), which was related to antigen processing and presentation and response to IFN-γ (Fig. 6D). Furthermore, neutrophils infiltrated and were expanded in the brain following HIE (Hyperacute, 8.57% vs 1%; Acute, 48.22% vs 0.79%; Subacute, 38.78% vs 2.63%). Mast cells increased during hyperacute and acute phases of HIE (Hyperacute, 30.71% vs 10.93%; Acute, 26.91% vs 13.02%), which was associated with mast cell activation, inflammation, proteoglycan metabolic process, and fibroblast migration (Fig. 6E). T cells and pDCs increased during acute and subacute phases of HIE. Two subtypes of B cells were discovered in our data. Significant DEGs of B cell-A type was related to regulation of B cell proliferation, formation of a pool of free 40S subunits, and regulation of calcium ion transmembrane transport, whereas B cell-B type was associated with metabolism of RNA, cell cycle, B cell activation. However, these two types of B cell merely increased at the subacute phase of HIE (A-type, 81.7% vs 5.4%; B-type, 80.13% vs 5.43%).

This study represents the first scRNA-seq investigation revealing cellular heterogeneity in neonatal HIE. In the hyperacute phase, astrocytes are activated in response to ROS, microglia engage in phagocytosis, and there is an increase in Stat3-induced ischemia-responsible OPCs and senescent LECs. In the acute phase, astrocytes contribute to inflammation and inhibit brain development, microglia proliferate, neuroblasts are affected by metal ions, and oligodendrocytes decrease. During the subacute phase, astrocytes facilitate tissue repair, and inflammatory microglia, induced by the IL27 and type I interferon pathways, expand with elevated MHC II expression. Metal ion-responsible neuroblasts and fibroblasts increase, while OCs further decrease. In addition, peripheral immune cells are involved throughout HIE. Neutrophils infiltrate and expand in the brain across all phases, Spp1^high macrophages, T cells, and pDCs increase during the acute and subacute phases, and B cells expand during the subacute phase.

Astrocytes play a critical role in synaptic development and function in the brain³⁵. HIE induces dynamic changes in astrocytes. During the hyperacute phase (3 h), astrocytes are activated in response to ischemia and hypoxia, exhibiting high expression of IEGs. These activated astrocytes facilitate translation processes. Notably, Igfbp4 emerged as a specific marker for these activated astrocytes. Igfbp4 encodes IGFBP4, which binds and inhibits IGF-I/II, with its levels regulated by PAPP-A dependent proteolysis³⁶. The IGF pathway is essential for organismal organism during development, and IGF1 knockout in mice leads to microcephaly³⁷. HIE-induced IGFBP4^high astrocytes may inhibit the brain development, and targeting IGFBP4 may be a potential therapy in the early stage. In addition, Wnt pathway mediated neural progenitor proliferation during developmental phase³⁸. Our results showed that acute HIE induced Sfrp1^highSfrp2^high astrocytes (A10). Sfrp1 and Sfrp2 are known for inhibiting Wnt pathway³⁹. Thus, HIE may result in growth-inhibitory astrocytes that impede both IGF and Wnt pathways during the acute phase. In the acute phase (2 d), astrocytes upregulated another IGF inhibitory factor-Igfbp5, suggesting ongoing inhibition of brain growth and development. In addition, astrocytes proliferated and contributed to neuroinflammation in the subacute of HIE. During the subacute phase (7 d), astrocytes are converted to repair phenotype, contributing to angiogenesis and axonal regeneration. Increased Agt expression in astrocytes during this phase may elevate blood pressure⁴⁰. Extracellular matrix (ECM) associated proteins expressing after injury hider the neurite growth, which was regulated by metalloproteinases (TIMPs). Timp-1 expressed in reactive astrocytes promotes ECM remodeling after brain injury⁴¹. Our data showed Timp4 was upregulated in HIE-induced astrocytes, indicating that Timp-4 may facilitate ECM remodeling.

During central nervous system (CNS) development, substantial remodeling occurs through synaptic pruning, leading to a precise neural circuit wiring diagram^42,43. Microglia contribute to synaptic plasticity by phagocytosing and eliminating excess dendrites, axons, and synapses generated during development^44,45. Our data indicate that phagocytic microglia are particularly prevalent in the early developmental stage (P7) and their numbers increase following HIE, likely to clear injured cell debris. HIE also induces microglial proliferation, though to a lesser extent than in other CNS injuries^20,46. However, a lower degree of microglial proliferation was induced via HIE compared to CNS injury. We speculate about the possible reasons including: i) microglia in the developmental stage owe proliferative characteristics and the proliferative potential is weak, while most adult microglia are homeostatic and the proliferative potential is stronger; ii) CNS injury broadly resulted in microglial death at the center sites, the absence of cells at the center sites allows the microglial proliferation for repopulation, whereas we found that microglial death or apoptosis did not appear during the hyperacute and acute phase of HIE. The specific mechanisms behind microglial proliferation in HIE require further investigation. Recent studies have shown that HDAC3 mediates proinflammatory microglial proliferation without affecting anti-inflammatory microglia, suggesting that further exploration of HDAC3 and other factors in HIE is warranted⁴⁷.

We identified microglial subtypes (M5, M8, M9, and M10) expressing markers associated with astrocytes, oligodendroglia, neuron, and OPC. Intriguingly, other studies have similarly observed microglial subsets with neuronal mRNA^48,49. It remains unclear whether these neuronal gene-enriched microglia endogenously express neuronal genes or phagocytose neurons containing such mRNA. The neuronal gene-enriched microglia increased after whisker deprivation⁴⁹. We found that neuronal gene-enriched microglia increase during the hyperacute phase of HIE, whereas phagocytic microglia are elevated at all stages of HIE. This implied that the increase of neuronal gene-enriched microglia did not occur with the increase of phagocytic microglia, suggesting that neuronal gene-enriched microglia may endogenously express neuronal genes rather than resulting from phagocytosis.However, the straight evidences need to be further obtained. Notably, MHCII^high inflammatory microglia, related to IFN I type signaling pathway, peak in the subacute phase of HIE. IFN-I-responsive microglia have been recognized recently, which engulf whole neurons and maintain excitatory/inhibitory balance during cortical development⁴⁹. Our scRNA-seq data also revealed that MHCII^high inflammatory microglia appear during brain development, which are largely boost during the subacute phase of HIE. These microglia may be derived from phagocytic microglia, which is supported by the expression of phagocytose related gene-Cd68. The findings that post-phagocytosis macrophages turn to the inflammatory phenotype also supports our opinion⁵⁰. Furthermore, our data showed that IL27-STAT1 signaling pathway may be vital for the generation of MHCII^high inflammatory microglia following HIE.

A previous study has identified two neuroblast populations in the developing mouse brain, one expressing Eomes, Tac2, and Calb2, and the other expressing Gal⁵¹. Our study expands this to twelve neuroblast populations, including a distinct subset associated with HIE, characterized by metal ion-related neurodegeneration and the expression of Nrgn, Clu, and Apoe⁵². Metal ion such as Fe, Cu, and Zn loading in the neuroblasts following HIE induced the generation of this population, which exhibited age-inappropriate neurodegeneration. However, detailed mechanism needs to be further explored. OPCs were yet affected at hyperacute stage of HIE, which were defined as ischemia responsible OPCs upregulating Vgf. The gene Vgf is NGF and BDNF inducible, encoding a 617 amino-acid precursor polypeptide. This precursor polypeptide is processed into some bioactive peptides that regulate neuronal survival and microglial phagocytose^53,54. However, the function of Vgf in OPCs following HIE deserve exploring. OCs exhibit a delayed response post-HIE, with antigen presentation-related OCs potentially inducing adaptive immunity and ER stress-related OCs leading to apoptosis⁵⁵.

Progressive acquired hydrocephalus of unknown origin was observed in HIE infants^56,57. Our data recognized a novel LEC associated with senescence following HIE, which may impair cerebrospinal fluid lymphatic drainage and contribute to hydrocephalus. Investigating anti-aging agents and the roles and mechanisms of these LECs could be beneficial. Fibroblast growth factors (FGFs), such as FGF-10, FGF-2, and FGF-21, are known to promote neurological recovery following HIE, but the role of fibroblasts in HIE remains unclear^58–60. Our findings indicate that fibroblasts are primarily altered during the subacute phase of HIE, potentially involved in brain remodeling.

Osteopontin (OPN), encoded by Spp1, is a cytokine-like glycoprotein that binds to integrins and CD44 and is upregulated in stroke, CNS injury, and neurodegeneration. It promotes inflammation and wound healing^61,62. During development, embryonic microglia participate axon tract and limit the size of the microcavities depending on the expression of OPN⁶³. In addition, some studies also defined Spp1⁺ macrophages as profibrotic macrophages that activates myofibroblasts⁶⁴. Our data reveal an expansion of Spp1⁺ macrophages and fibroblasts following HIE, highlighting the need for further understanding of their roles and mechanisms.

Conflict of Interest

The authors declare no interests.

Author Contribution

J.W. designed this study. X.C. conducted sample collection and scRNA-seq analysis. J.W. established HIE models. X.C. and X.T. helped with the scRNA-seq analysis. X.C. wrote the manuscript. J.W. modified the manuscript.

Acknowledgement

This study was supported in part by grants from the Social Development Science and Technology Project of Wenling City (No. 2023S00150), the Postdoctoral Fellowship Program of CPSF under Grant Number GZC20232299, the National Natural Science Foundation of China (No. 82401602).

Data Availability

The raw data files of scRNA-seq associated with this study have been deposited in the Gene Expression Omnibus under the number GSE273044. Additional data can be requested from the corresponding author.

Peeples ES, Genaro-Mattos TC, Ferroptosis. A Promising Therapeutic Target for Neonatal Hypoxic-Ischemic Brain Injury. Int J Mol Sci 23(2022).
Yang M, Wang K, Liu B, Shen Y, Liu G. Hypoxic-Ischemic Encephalopathy: Pathogenesis and Promising Therapies. Mol Neurobiol (2024).
Ristovska S, Stomnaroska O, Danilovski D. Hypoxic Ischemic Encephalopathy (HIE) in Term and Preterm Infants. Pril (Makedon Akad Nauk Umet Odd Med Nauki). 2022;43:77–84.
Korf JM, McCullough LD, Caretti V. A narrative review on treatment strategies for neonatal hypoxic ischemic encephalopathy. Transl Pediatr. 2023;12:1552–71.
Neves D, Salazar IL, Almeida RD, Silva RM. Molecular mechanisms of ischemia and glutamate excitotoxicity. Life Sci. 2023;328:121814.
Nuñez A et al. [Oxidative stress in perinatal asphyxia and hypoxic-ischaemic encephalopathy]. An Pediatr (Engl Ed) 88, 228.e1-228.e9 (2018).
Zhao M et al. Oxidative Stress in Hypoxic-Ischemic Encephalopathy: Molecular Mechanisms and Therapeutic Strategies. Int J Mol Sci 17(2016).
Zhou T, et al. Mild hypothermia alleviates oxygen-glucose deprivation/reperfusion-induced apoptosis by inhibiting ROS generation, improving mitochondrial dysfunction and regulating DNA damage repair pathway in PC12 cells. Apoptosis. 2023;28:447–57.
Mülling K, et al. Neutrophil dynamics, plasticity and function in acute neurodegeneration following neonatal hypoxia-ischemia. Brain Behav Immun. 2021;92:234–44.
Liu F, McCullough LD. Inflammatory responses in hypoxic ischemic encephalopathy. Acta Pharmacol Sin. 2013;34:1121–30.
He N, et al. Resveratrol suppresses microglial activation and promotes functional recovery of traumatic spinal cord via improving intestinal microbiota. Pharmacol Res. 2022;183:106377.
Juul SE, Ferriero DM. Pharmacologic neuroprotective strategies in neonatal brain injury. Clin Perinatol 41, 119 – 31 (2014).
Molloy EJ, et al. Neuroprotective therapies in the NICU in term infants: present and future. Pediatr Res. 2023;93:1819–27.
Douglas-Escobar M, Weiss MD. Hypoxic-ischemic encephalopathy: a review for the clinician. JAMA Pediatr. 2015;169:397–403.
Variane GFT, et al. Remote Monitoring for Seizures During Therapeutic Hypothermia in Neonates With Hypoxic-Ischemic Encephalopathy. JAMA Netw Open. 2023;6:e2343429.
Shankaran S, et al. Whole-body hypothermia for neonates with hypoxic-ischemic encephalopathy. N Engl J Med. 2005;353:1574–84.
Laptook AR, et al. Effect of Therapeutic Hypothermia Initiated After 6 Hours of Age on Death or Disability Among Newborns With Hypoxic-Ischemic Encephalopathy: A Randomized Clinical Trial. JAMA. 2017;318:1550–60.
Shankaran S, et al. Effect of depth and duration of cooling on deaths in the NICU among neonates with hypoxic ischemic encephalopathy: a randomized clinical trial. JAMA. 2014;312:2629–39.
Wang J, et al. Stem cell-derived mitochondria transplantation: a novel strategy and the challenges for the treatment of tissue injury. Stem Cell Res Ther. 2018;9:106.
Wang J, et al. Single-cell transcriptome analysis reveals the immune heterogeneity and the repopulation of microglia by Hif1α in mice after spinal cord injury. Cell Death Dis. 2022;13:432.
Zhong J, et al. Single-cell RNA sequencing analysis reveals the relationship of bone marrow and osteopenia in STZ-induced type 1 diabetic mice. J Adv Res. 2022;41:145–58.
Wälchli T, et al. Single-cell atlas of the human brain vasculature across development, adulthood and disease. Nature. 2024;632:603–13.
Arneson D, et al. Single cell molecular alterations reveal target cells and pathways of concussive brain injury. Nat Commun. 2018;9:3894.
Zheng K, et al. Single-cell RNA-seq reveals the transcriptional landscape in ischemic stroke. J Cereb Blood Flow Metab. 2022;42:56–73.
Depp C, et al. Myelin dysfunction drives amyloid-β deposition in models of Alzheimer's disease. Nature. 2023;618:349–57.
Absinta M, et al. A lymphocyte-microglia-astrocyte axis in chronic active multiple sclerosis. Nature. 2021;597:709–14.
Julien P, et al. Synergistic effect of sildenafil combined with controlled hypothermia to alleviate microglial activation after neonatal hypoxia-ischemia in rats. J Neuroinflammation. 2024;21:31.
Stuart T, et al. Comprehensive Integration of Single-Cell Data. Cell. 2019;177:1888–e190221.
Korsunsky I, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods. 2019;16:1289–96.
Zhou Y, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10:1523.
Milich LM et al. Single-cell analysis of the cellular heterogeneity and interactions in the injured mouse spinal cord. J Exp Med 218(2021).
Hammond TR, et al. Single-Cell RNA Sequencing of Microglia throughout the Mouse Lifespan and in the Injured Brain Reveals Complex Cell-State Changes. Immunity. 2019;50:253–e2716.
Unger MS, et al. Doublecortin expression in CD8 + T-cells and microglia at sites of amyloid-β plaques: A potential role in shaping plaque pathology? Alzheimers Dement. 2018;14:1022–37.
Kumar T, et al. A spatially resolved single-cell genomic atlas of the adult human breast. Nature. 2023;620:181–91.
Zhou B, Zuo YX, Jiang RT. Astrocyte morphology: Diversity, plasticity, and role in neurological diseases. CNS Neurosci Ther. 2019;25:665–73.
Qin X, Byun D, Lau KH, Baylink DJ, Mohan S. Evidence that the interaction between insulin-like growth factor (IGF)-II and IGF binding protein (IGFBP)-4 is essential for the action of the IGF-II-dependent IGFBP-4 protease. Arch Biochem Biophys. 2000;379:209–16.
Rusin D et al. Microglia-Derived Insulin-like Growth Factor 1 Is Critical for Neurodevelopment. Cells 13(2024).
Singh KK et al. Common DISC1 polymorphisms disrupt Wnt/GSK3β signaling and brain development. Neuron 72, 545 – 58 (2011).
Rodriguez J, et al. SFRP1 regulates the growth of retinal ganglion cell axons through the Fz2 receptor. Nat Neurosci. 2005;8:1301–9.
Morimoto S, et al. Elevated blood pressure in transgenic mice with brain-specific expression of human angiotensinogen driven by the glial fibrillary acidic protein promoter. Circ Res. 2001;89:365–72.
Jaworski DM. Differential regulation of tissue inhibitor of metalloproteinase mRNA expression in response to intracranial injury. Glia. 2000;30:199–208.
Riccomagno MM, Kolodkin AL. Sculpting neural circuits by axon and dendrite pruning. Annu Rev Cell Dev Biol. 2015;31:779–805.
Schuldiner O, Yaron A. Mechanisms of developmental neurite pruning. Cell Mol Life Sci 72, 101 – 19 (2015).
Faust TE, Gunner G, Schafer DP. Mechanisms governing activity-dependent synaptic pruning in the developing mammalian CNS. Nat Rev Neurosci. 2021;22:657–73.
Luo L, O'Leary DD. Axon retraction and degeneration in development and disease. Annu Rev Neurosci. 2005;28:127–56.
Bellver-Landete V, et al. Microglia are an essential component of the neuroprotective scar that forms after spinal cord injury. Nat Commun. 2019;10:518.
Zhang Y, et al. Arresting the bad seed: HDAC3 regulates proliferation of different microglia after ischemic stroke. Sci Adv. 2024;10:eade6900.
Li Y, et al. Decoding the temporal and regional specification of microglia in the developing human brain. Cell Stem Cell. 2022;29:620–e6346.
Escoubas CC et al. Type-I-interferon-responsive microglia shape cortical development and behavior. Cell 187, 1936–1954.e24 (2024).
Wang X, et al. Macrophages in spinal cord injury: phenotypic and functional change from exposure to myelin debris. Glia. 2015;63:635–51.
Hochgerner H, Zeisel A, Lönnerberg P, Linnarsson S. Conserved properties of dentate gyrus neurogenesis across postnatal development revealed by single-cell RNA sequencing. Nat Neurosci. 2018;21:290–9.
Marsh TW et al. A genetic and proteomic comparison of key AD biomarkers across tissues. Alzheimers Dement (2024).
El Gaamouch F, et al. VGF-derived peptide TLQP-21 modulates microglial function through C3aR1 signaling pathways and reduces neuropathology in 5xFAD mice. Mol Neurodegener. 2020;15:4.
Beckmann ND, et al. Multiscale causal networks identify VGF as a key regulator of Alzheimer's disease. Nat Commun. 2020;11:3942.
Wang J, et al. Local Delivery of β-Elemene Improves Locomotor Functional Recovery by Alleviating Endoplasmic Reticulum Stress and Reducing Neuronal Apoptosis in Rats with Spinal Cord Injury. Cell Physiol Biochem. 2018;49:595–609.
Chiyoda H, Kobayashi S, Yokoi K, Iwata O, Katano H. Acquired hydrocephalus following hypoxic ischemic encephalopathy without intraventricular hemorrhage: A case report. J Neonatal Perinat Med. 2023;16:569–71.
Chao CP, Zaleski CG, Patton AC. Neonatal hypoxic-ischemic encephalopathy: multimodality imaging findings. Radiographics. 2006;26(Suppl 1):S159–72.
Fang M, et al. Protective effects of FGF10 on neurovascular unit in a rat model of neonatal hypoxic-ischemic brain injury. Exp Neurol. 2020;332:113393.
Herrmann JR, et al. FGF21 modulates hippocampal cold-shock proteins and CA2-subregion proteins in neonatal mice with hypoxia-ischemia. Pediatr Res. 2023;94:1355–64.
Ye L, et al. FGF21 promotes functional recovery after hypoxic-ischemic brain injury in neonatal rats by activating the PI3K/Akt signaling pathway via FGFR1/β-klotho. Exp Neurol. 2019;317:34–50.
Schroeter M, Zickler P, Denhardt DT, Hartung HP, Jander S. Increased thalamic neurodegeneration following ischaemic cortical stroke in osteopontin-deficient mice. Brain. 2006;129:1426–37.
Ellison JA, Barone FC, Feuerstein GZ. Matrix remodeling after stroke. De novo expression of matrix proteins and integrin receptors. Ann N Y Acad Sci. 1999;890:204–22.
Lawrence AR, et al. Microglia maintain structural integrity during fetal brain morphogenesis. Cell. 2024;187:962–e98019.
Ouyang Q, et al. Depleting profibrotic macrophages using bioactivated in vivo assembly peptides ameliorates kidney fibrosis. Cell Mol Immunol. 2024;21:826–41.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Single-cell Transcriptomic Profiling of Brains in Newborn Rats Following Hypoxic Ischemic Encephalopathy

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Methods

Animals

Models of HIE

Single cell dissociation

Single cell RNA sequencing

Results

Single-cell profiling of the brain after HIE

Dynamic changes of astrocytes following HIE

Dynamic changes of microglia following HIE

Dynamic changes of neuroblast and oligodendrocyte lineage following HIE

Dynamic changes of endotheliocytes and fibroblasts following HIE

Dynamic changes of infiltrated immune cells following HIE

Discussion

Declarations

Conflict of Interest

Author Contribution

Acknowledgement

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1