Abstract

Objective

   Intracerebral hemorrhage (ICH) is a subtype of stroke with high
   mortality and morbidity rates. Our aim was to comprehensively analyze
   transcriptome and proteome in an experimental ICH model.

Methods

   All mice were divided into ICH model (n = 3) and sham groups (n = 3).
   ICH was induced by collagenase VII. The ipsilateral hemisphere was used
   for whole transcriptome and proteomics resequencing. After
   preprocessing, differentially expressed lncRNAs (DElncRNAs), mRNAs
   (DEmRNAs), miRNAs (DEmiRNAs), and DEproteins between ICH and sham
   groups were identified. Functional enrichment analysis was performed
   using the clusterProfiler package, followed by protein–protein
   interaction (PPI) analysis. After that, the Pearson correlation
   coefficient between DEmRNAs and DElncRNAs or between DEmRNAs and
   DEproteins was calculated. DElncRNAs with similar functions were
   analyzed by the GOSemSim package. After prediction of DEmiRNA–DEmRNA
   and DElncRNA–DEmiRNA relationships, a competing endogenous RNA (ceRNA)
   network was constructed. Several DEmRNAs and DElncRNAs were validated
   in ipsilateral hemisphere tissues of the ICH model and control groups
   using RT-qPCR and western blot.

Results

   Between the ICH and sham groups, 31 DElncRNAs, 367 DEmRNAs, 35
   DEmiRNAs, and 96 DEproteins were identified. DEmRNAs were mainly
   enriched in inflammation, such as cytokine–cytokine receptor
   interaction, IL-17, and TNF signaling pathways. A PPI network of
   DEmRNAs was constructed and hub genes were identified, such as IL6
   (degree = 59), TNF (degree = 44), and CXCR2 (degree = 39). 24 DElncRNAs
   with similar functions were identified, including 15 up- and 9
   down-regulated lncRNAs. After integration of DEmiRNA–DEmRNA and
   DElncRNA–DEmiRNA relationships, we constructed a ceRNA network,
   composed of 71 DEmRNAs, 17 DEmiRNAs, and 12 DElncRNAs. RT-qPCR and
   western blot results confirmed that C3, Fga, and Slc4a1 proteins were
   more lowly expressed and Penk was more highly expressed in ICH than
   control groups, which could become potential markers for ICH.

Conclusion

   Our findings identified ICH-related DE-RNAs and proteins and potential
   molecular mechanisms of ICH by transcriptome resequencing and
   quantitative proteomic analyses.

   Keywords: intracerebral hemorrhage, transcriptome resequencing,
   proteomic analyses, inflammation, competing endogenous RNA,
   protein–protein interactions

Introduction

   Strokes are divided into either ischemic stroke or ICH. ICH accounts
   for about 15% to 20% of all stroke cases, characterized by hematoma
   expansion and inflammation. ICH patients have a mortality rate of up to
   40% in the first month. The mortality and disability rates of ICH
   patients is much higher than that of ischemic stroke patients ([37]Fang
   et al., 2013). Although efforts have been made to reduce ICH and
   post-ICH complications, the clinical outcomes have been suboptimal.
   About 20% of ICH survival patients suffer from neurological dysfunction
   ([38]Duan et al., 2016). However, effective treatment options for ICH
   are still lacking, and few studies provide evidence to guide ICH
   treatment ([39]Jia et al., 2018). ICH patients’ poor prognosis is
   closely related to the complicated pathogenesis of ICH. A previous
   study has shown that targeting ICH-related molecules could be a
   promising therapeutic strategy ([40]Liu et al., 2019). Thus, it is
   urgent to explore and understand the molecular mechanisms of ICH.

   Non-coding RNAs (ncRNAs), such as lncRNA and miRNA, exhibit important
   biological functions ([41]Beermann et al., 2016; [42]Matsui and Corey,
   2017). ncRNAs could affect the expression of target genes. lncRNA, a
   non-coding RNA larger than 200 nucleotides, has been reported to play
   an important role in the pathophysiology of ICH ([43]Zhang and Wang,
   2019). Furthermore, lncRNA as a sponge of miRNA can indirectly regulate
   the expression of downstream messenger RNA (mRNA), which is described
   as ceRNA ([44]Park et al., 2018). The interactions between genes or
   proteins are involved in the pathogenesis of diseases. The functions of
   mRNA, miRNA, lncRNA, and proteins in ICH remains largely unknown.

   Herein, this study comprehensively analyzed the transcriptome and
   proteome of ICH based on collagenase VII-induced ICH mouse models,
   which could provide an insight into ICH-related DE-RNAs, proteins, and
   potential molecular mechanisms.

Materials and Methods

Animals

   Male C57BL/6 mice (age: 10–12 weeks; weight: 22–25 g) were purchased
   from the Animal Institute of the Third Military Medical University. All
   mice were fed under a 12 h light/dark cycle in a temperature-controlled
   and specific pathogen-free environment. All animal experiments
   conformed to the animal experiment manual approved by the Animal Ethics
   Committee of Zunyi Medical University. All experiments were performed
   and reported according to the Animal Research: Reporting in vivo
   Experiments (ARRIVE) guidelines.

ICH Mouse Model

   The collagenase VII-induced ICH model was established as previously
   reported ([45]Xie et al., 2016). All mice were randomly divided into
   ICH model group (n = 3) and sham operation (n = 3). All mice were
   anesthetized by intraperitoneal injection of pentobarbital sodium and
   placed on a brain stereotaxic apparatus (RWD, China) in the
   ventricumbent position. A previously reported coordinate point
   (coordinates: 0.2 mm anterior, 2.3 mm lateral, and 3.5 mm depth to
   bregma) was utilized to inject 1 μL bacterial collagenase (0.0375 units
   per 1 μL, type VII-S; Sigma-Aldrich, United States) into the striatum
   at a rate of approximately 0.1 μL/min using a microinjection pump
   (Longer, TJ-2A/L0107-2A, China). After that, the needle was kept at the
   injection point for 5 min to prevent liquid backflow. The microsyringe
   was removed slowly and the cranial pinhole was closed with bone wax.
   The incisions in the skin were sutured. The sham operation group was
   injected with an equivalent volume of PBS only, and the other
   operations were the same.

Mice Hemisphere Harvest

   In this study, all mice were scanned by a Bruker 7T MRI (70/20) system
   (BrukerBiospin, Billerica, MA, United States) ([46]Cao et al., 2016).
   After the rotarod test pre- and post-CCI, mice were anesthetized with
   gas mixture (induction: 5% isoflurane with 1 L/min O[2], maintenance:
   1% isoflurane with 1 L/min O[2]), mounted in a Bruker animal bed, and
   their body temperature was maintained at 37°C with respiratory rate
   continuously monitored. T2-weighted images were acquired using RARE (TR
   = 4000, TE = 45, RARE factor 8, 0.5 mm, FOV 2.5 cm, 256 × 256). Images
   were analyzed using Bruker ParaVision 6.0 software. The lesion volumes
   were determined as pixels that had T2 values higher than the mean plus
   two standard deviations of the value in the homologous contralesional
   region. The model was confirmed to be successfully constructed before
   euthanizing at 24 h after ICH injury or sham injury, and the whole
   brain was divided into two halves, as described in a previous study
   ([47]Cao et al., 2016). The ipsilateral hemisphere was used for whole
   transcriptome resequencing by Novaseq 6000 (Illumina, United States)
   and whole proteomics resequencing that were analyzed on an Orbitrap
   Fusion Lumos Tribrid mass spectrometer (Thermo Scientific, United
   States).

Whole Transcriptome Resequencing Analysis

   Total RNA was extracted from tissues using TRIzol reagent (Invitrogen,
   United States). Standard denaturing gel electrophoresis was used to
   assess RNA integrity. Extracted RNA was transcribed into cDNA. Mouse
   reference genome and annotation information were downloaded using
   Ensembl Genome Browser [version: GRCm38.p6 (GCA_000001635.8)]. An index
   of the reference genome was created via hisat2. The transcriptome
   sequencing double-end data were firstly cleaned using Trim Galore. Trim
   Galore can automatically identify and remove the 3′ end adapter. In
   this way, the transcriptome data were quantified after obtaining clean
   data from raw data. Two separate library preparations were used for the
   long and small RNAs. The ribosomal depletion was utilized for library
   preparations of the long RNAs. Furthermore, small RNA-seq cDNA library
   preparation was performed for the small RNAs.

   The quantification of the RNA (including lncRNA and mRNA) transcriptome
   was performed using hisat2. The parameters were defaulted. The
   alignment rates were as follows: M1: 96.65% overall alignment rate, M2:
   96.61% overall alignment rate, M3: 96.63% overall alignment rate, M7:
   96.47% overall alignment rate, M8: 96.86% overall alignment rate, and
   M9: 96.52% overall alignment rate. Among them, M1-3 represented three
   ICH mouse models and M7-9 represented three sham operation mice. After
   sorting the mapping results by samtools, featureCounts was used to
   quantify. The quantitative results of the RNA transcriptome data were
   obtained for further analysis.

   The miRNA transcriptome was quantified using miRDeep2. All mouse miRNA
   sequence data were downloaded as alignment references and as annotated