Transcriptomic Hallmarks of Ischemia-Reperfusion Injury

Ischemia reperfusion injury (IRI) is associated with a broad array of life-threatening medical conditions including myocardial infarct, cerebral stroke, and organ transplant. Although the pathobiology and clinical manifestations of IRI are well reviewed by previous publications, IRI-related transcriptomic alterations are less studied. This study aimed to reveal a transcriptomic hallmark for IRI by using the RNA-sequencing data provided by several studies on non-human preclinical experimental models. In this regard, we focused on the transcriptional responses of IRI in an acute time-point up to 48 h. We compiled a list of highly reported genes in the current literature that are affected in the context of IRI. We conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses and found many of the up-regulated genes to be involved in cell survival, cell surface signaling, response to oxidative stress, and inflammatory response, while down-regulated genes were predominantly involved in ion transport. Furthermore, by GO analysis, we found that multiple inflammatory and stress response processes were affected after IRI. Tumor necrosis factor alpha (TNF) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling pathways were also highlighted in the Kyoto Encyclopedia of Genes and Genomes enrichment analysis. In the last section, we discuss the treatment approaches and their efficacy for IRI by comparing RNA sequencing data from therapeutic interventions with the results of our cross-comparison of differentially expressed genes and pathways across IRI.


Introduction
Ischemia reperfusion injury (IRI) is imposed by the transient loss of circulation to a specific tissue for various durations of time, followed by the restoration of blood flow that results in widespread inflammatory responses causing secondary injury [1]. IRI commonly occurs in myocardial infarction [2], cerebral stroke [3], and cases of transplantation of the kidney [4] and liver [5]. Across these injury modalities, loss of circulation causes ischemic hypoxia, which has vast effects in cell metabolism, ionic balance maintenance, and mitochondrial function [1].
In hypoxic conditions, glycolysis becomes the sole energy-producing process and the mitochondrial electron transport system is inhibited, resulting in buildup of glycolytic by-products and reactive oxidative species (ROS) [6]. This disrupts ionic balances, thereby disrupting transmembrane potential. In the mitochondria, a lack of oxygen causing buildup of hydrogen ions across the membrane may cause the mitochondrial permeability transition pore to open, decreasing transmembrane potential and ultimately resulting in bursting and subsequent release of apoptogens and organelle components, contributing to damageassociated molecular patterns [6]. Similarly, buildup of ROS caused by mitochondrial dysfunction impairs protein folding in the endoplasmic reticulum [7] and activate protein complexes, such as mitogen-activated protein kinase (MAPK) [8], calcium/calmodulindependent protein kinase (CaMK) [9], protein kinase C [10], and receptor-interacting protein kinases [11], which initiate several pro-inflammatory cascades. Ionic imbalances inflict similar damage across the cell membrane, where voltage-gated channels respond to the absence of oxygen by increasing ionic exchange to maintain tonicity, eventually leading to excess intracellular calcium and cytotoxicity [12]. In cases of stroke, the dramatic release of glutamate from calcium-saturated cells causes widespread neuronal glutamatergic excitotoxicity and cell death, and inhibiting sodium channels with was found to attenuate this cascade [13].
Ischemia and ischemia reperfusion effects were also observed on the genomic level, where ROS were found to drive epigenetic changes in expression of mRNA [14] and noncoding RNAs, facilitating an array of further cellular responses [1]. While the molecular footprint of IRI was extensively studied and articulated, transcriptional responses have only recently garnered attention and lack any in-depth characterization. Moreover, these transcriptional events are likely to play a far more significant role than is currently appreciated, and understanding them is critical to addressing the pathophysiology of IRI [14].

Systematic Review
The PubMed database (https://pubmed.ncbi.nlm.nih.gov/, accessed on 14 April 2021) was thoroughly searched from inception to April 2021. Three searches were conducted using the following key words in either the abstract or the title: ("IRI" AND "RNAsequencing"), ("ischemia reperfusion injury" AND "RNA-sequencing"), and ("ischemia reperfusion injury" AND "transcriptomic"). The selected articles had to contain open access datasets with the full list of differentially expressed genes (DEGs) from bulk RNAsequencing. Studies on single-cell RNA-sequencing were excluded. The changes in the mRNA profile were extracted for the acute phase (from the injury onset up to 48 h) post-IRI. Selected articles also had to contain either naïve control or uninjured sham in order to compare the normal gene expression profile and the IRI-related changes in the mRNA profile. The studies reporting gene expression changes only after applying a treatment for IRI were excluded.

Data Extraction and Handling
We extracted the DEGs (adjusted p-value < 0.05) from each publication and crossreferenced them based on the reported gene symbols. The resultant list included genes reported with varying frequencies (scores) across studies ranging from 7 to 1. Next, this list was filtered for consistency. We kept only the genes that had consistent expression directionality in at least 75% of the reports for that gene. We then divided these genes into up-and down-regulated lists. Finally, we ranked these genes based on the number of times they were reported across studies. The genes with 4 or more reports (out of 7) were identified as "commonly reported DEGs" (Tables S4 and S5), which were used for further enrichment analysis. The highest ranked items of up-and down-regulated lists were represented as "highly reported DEGs" in Tables 2 and 3. All the initial data handling was performed on Excel version 2016 (Microsoft; WA, USA). Table 2. Highly repeated up-regulated genes in IRI.

Itga5
Programmed cell death Atf3 Cell proliferation, programmed cell death Cyr61 Cell proliferation, programmed cell death Litaf Programmed cell death Stat3 Programmed cell death, inflammatory response Icam1 Inflammatory response, programmed cell death Sik1 Programmed cell death Tlr2 Inflammatory response Zc3h12a Inflammatory response Epha2 Inflammatory response Maff Response to stress Nes Cell proliferation, programmed cell death Hmox1 Response to stress, inflammatory response Myc Programmed cell death, response to stress Cd44 Inflammatory response Hbegf Cell proliferation Ifrd1 Developmental process Pdpn Cell proliferation, programmed cell death Cd14 Inflammatory response Jun Cell proliferation, programmed cell death Csf1 Inflammatory response Gadd45a Programmed cell death, response to stress Ccl2 Inflammatory response Klf6 Developmental process Hspb1 Response to stress, programmed cell death Olr1 Inflammatory response Mafk Response to stress Zyx Inflammatory response Osmr Inflammatory response Tifa Response to stress Bcl3 Inflammatory response, programmed cell death Irf1 Cell proliferation, programmed cell death Tnfrsf1a Inflammatory response Cxcl1 Inflammatory response Txnrd1 Cell proliferation, response to stress Spon2 Cell surface adhesion and signaling, response to stress Myd88 Inflammatory response Bach1 Response to stress Table 3. Highly repeated down-regulated genes in IRI.

Cabp1
Voltage-gated calcium ion channel regulator Pclo Presynaptic cytoskeletal matrix Npy1r Synaptic signaling Tenm2 Formation of growth cone in neural cells Cbx7 Cellular lifespan regulator Camk4 Transcriptional regulation Kcng1 Voltage-gated potassium (Kv) channels subunit L1cam Cell surface adhesion and signaling Dlg2 Response to oxidative stress and DNA damage Scn4b Sodium channel beta subunit Kcnab1 Cytoplasmic potassium channel subunit Rasgrp1 T cell/B cell regulator Kcnt1 Potassium-sodium activated channel subunit Adora2a Cell surface adhesion and signaling Phactr1 Endothelial cell survival Lzts3 Maturation of dendritic spines Stard8 Cell surface adhesion and signaling Agrn Cytoplasmic Ca ion regulator Cacnb3 Voltage-dependent calcium channel regulatory subunit Cobl Reorganization of the actin cytoskeleton

GGO Enrichment Analysis
The GO categories "Molecular Function (MF)", "Biological Process (BP)", and "Cellular Components (CC)" were analyzed for significantly enriched terms using the PANTHER enrichment test [32]. The analysis was run against the Mus musculus reference list using Fisher's exact test and p-values were corrected by false discovery rate (FDR) calculation. Finally, all terms with an FDR < 0.05 were collected and the top 20 terms for each category were graphed using the ggplot2 package in R [33].

KEGG Analysis
The most highly affected KEGG pathways were isolated using the clusterProfiler package in R [33,34]. Commonly repeated DEGs were enriched with KEGG dataset for the Mus musculus organism (organism code: mmu) and pathways with an adjusted p-value of < 0.05 were plotted. Finally, the top 20 significant KEGG pathways were graphed on bubble charts using the ggplot2 package in R.

Venn Diagram
We graphed the DEG lists from all studied publications (genes with adjusted p-values of <0.05) using the Venn diagrams drawer online tool (http://bioinformatics.psb.ugent. be/webtools/Venn/, accessed on 20 June 2021). To analyze organ-specific changes, we combined the DEG lists from studies on cerebral models (three publications) and used this in our comparison across organ models (spinal cord, hepatic, renal, and myocardial; one publication per organ).

Transcriptomic Hallmark of Ischemic Reperfusion Injury
The injurious effects of reperfusion injury appear over hours and days following ischemia. In this article, we first aimed to outline the transcriptomic hallmark of IRI occurring over the 48-h window post-ischemia. We did this by conducting a comparison analysis between the DEGs of mRNA sequencing data provided by multiple studies on non-human preclinical experimental models (full list is provided in Table S1). In our comparative analysis, we only considered the genes that were consistently reported to be either up-or down-regulated. The genes with the highest consistency score are presented as "highly repeated DEGs" in Tables 2 and 3. Among the upregulated genes, there are multiple inflammatory response genes, including Bcl3, Tnfrsf1a, Cxcl1, and Tlr2. Furthermore, the upregulation of Litaf, Myc, Stat3, Atf3, and Cyr61 suggested the involvement of cell survival and apoptotic pathways. Other contributing processes may be cell surface adhesion and signaling (Icam1, Spon2, and Itga5) and oxidative stress response (Hspb1 and Hmox1). Interestingly, most of the consistently downregulated genes related to ion channels (e.g., Scn4b, Cabp1, Kcnab1, Kcnt1, Kcng1, and Cacnb3). Other downregulated genes play roles in cytoskeleton regulation (Pclo and Cobl) and response to hypoxia and DNA damage (Dlg2).
Although the selected articles for comparative analysis performed bulk RNA sequencing, we noticed several of our highly repeated DEGs matched with single-cell RNA sequencing results from Zamanian et al., 2012. In this study, the authors isolated astrocytes from an ischemic condition, and reported differential expression patterns [35]. The results suggested that astrocytes from the ischemic condition displayed a protective phenotype with a specific gene expression pattern from other types of injuries, such as neuroinflammation. Notably, some of our highly repeated DEGs were similar to the reported genes for these protective astrocytes from the ischemic condition (Litaf, Bcl3, Hmox1, Myc, Stat3, Atf3, Icam1, Tnfrsf1a, Cxcl1, and Hspb1). This information supports the notion that astrocytes may be one of the key contributors to IRI and likely play a protective role in this context.

Pathways Involved in Ischemia Reperfusion Injury
Next, we used a list of "commonly reported DEGs" in IRI conditions (4 out of 7) to run functional enrichment analysis. This list consisted of 571 upregulated and 135 downregulated genes (Tables S4 and S5). GO enrichment analysis was used to study biological processes, cellular components, and molecular functions implicated in IRI. The top 20 up-and down-regulated terms are reported in Figures 1 and 2, respectively. Among the most highly enriched biological processes, several upregulated DEGs related to immune signaling and cell death regulation, while downregulated DEGs were largely involved in intercellular signaling and synaptic transmission. The upregulated genes related to molecular functions primarily involved binding proteins, while downregulated genes related to transport channels. Finally, among enriched cellular component DEGs, those upregulated were found to relate to the cell periphery (GO:0071944), while those composing the synapse (GO:0045202) were downregulated.
Further, we studied the engaged pathways using KEGG enrichment analysis ( Figure 3). The pathway analysis for upregulated genes highlighted multiple inflammatory response pathways, such as TNF, NF-κB, and IL-12 signaling pathways, as well as cytokine-cytokine receptor interaction, and toll-like receptor signaling. Conversely, the top dysregulated pathways involving significantly downregulated genes included oxytocin and calcium signaling pathways and long-term potentiation.

Organ-Specific Differential Expression in Response to IRI
To illustrate transcriptomic changes in each organ, we used a Venn diagram of DEGs from the selected studies (see Table 1) categorized by organ (details are provided in Table S2). We found 38 DEGs common across all five organs (Figure 4). Among these were Bcl3, Icam1, Itag5, Atf3, and Myc, which are involved in apoptosis and cell death. We also observed transcriptomic changes in genes involved in the inflammatory response, including Stat3, Tlr2, Cxcl1, Epha2, and Myd88. Furthermore, based on these 38 genes, TNF, NF-κB, and MAPK signaling pathways appeared to be highly implicated in IRI response.    Table S9-S11).
Further, we studied the engaged pathways using KEGG enrichment analysis ( 3). The pathway analysis for upregulated genes highlighted multiple inflammato sponse pathways, such as TNF, NF-κB, and IL-12 signaling pathways, as well as cyt cytokine receptor interaction, and toll-like receptor signaling. Conversely, the top d ulated pathways involving significantly downregulated genes included oxytocin an cium signaling pathways and long-term potentiation.   Table S12 and S13).

Organ-Specific Differential Expression in Response to IRI
To illustrate transcriptomic changes in each organ, we used a Venn diagram of DEGs from the selected studies (see Table 1) categorized by organ (details are provided in Table  S2). We found 38 DEGs common across all five organs (Figure 4). Among these were Bcl3, Icam1, Itag5, Atf3, and Myc, which are involved in apoptosis and cell death. We also observed transcriptomic changes in genes involved in the inflammatory response, including Stat3, Tlr2, Cxcl1, Epha2, and Myd88. Furthermore, based on these 38 genes, TNF, NF-B, and MAPK signaling pathways appeared to be highly implicated in IRI response.   Table S2).

Therapeutic Interventions Act on Genes and Pathways Involved in IRI
Therapeutic interventions were investigated for their efficacy at attenuating IRI. These have included ischemic pre-and post-conditioning and delivery of stem cells to the affected organs. It is likely that treatment approaches that achieve beneficial effects act on the same genes and pathways implicated in IRI via their counter-regulation. Here, we will compare RNAseq data from therapeutic interventions with the results of our cross-comparison of DEGs and pathways across IRI.

Ischemic Postconditioning
Zhang et al. (2019) used ischemic postconditioning (IPO) of the liver in a wildtype mouse model, whereby, following 1 h of ischemia, the porta hepatis was re-perfused for three 5 s cycles prior to full reperfusion for four hours [18]. The extent of injury was then compared between animals that received IPO and those that only underwent IRI. Among those we found to be implicated across IRI conditions, the authors found Cyr61 to be downregulated by IPO. Cyr61 belongs to the CCN protein family and regulates cell adhesion, proliferation, and apoptosis [36,37]. Additionally, they observed significant upregulation of Atf3 in IPO, the product of which is a member of the ATF/cAMP-responsive element binding protein transcription factor family and has been found to repress macrophage pro-inflammatory gene expression in IRI [38,39]. Several pathways the authors found to be differentially regulated in liver IPO were similar to those we found to be highly implicated across IRI studies. IPO significantly downregulated the MAPK and IL-17 signaling pathways, which modulate cell proliferation and apoptosis, and inflammatory cytokines and apoptosis, respectively, suggesting their role as potential protective mechanisms in IRI [18].  Table S2).

Therapeutic Interventions Act on Genes and Pathways Involved in IRI
Therapeutic interventions were investigated for their efficacy at attenuating IRI. These have included ischemic pre-and post-conditioning and delivery of stem cells to the affected organs. It is likely that treatment approaches that achieve beneficial effects act on the same genes and pathways implicated in IRI via their counter-regulation. Here, we will compare RNAseq data from therapeutic interventions with the results of our cross-comparison of DEGs and pathways across IRI.

Ischemic Postconditioning
Zhang et al. (2019) used ischemic postconditioning (IPO) of the liver in a wildtype mouse model, whereby, following 1 h of ischemia, the porta hepatis was re-perfused for three 5 s cycles prior to full reperfusion for four hours [18]. The extent of injury was then compared between animals that received IPO and those that only underwent IRI. Among those we found to be implicated across IRI conditions, the authors found Cyr61 to be downregulated by IPO. Cyr61 belongs to the CCN protein family and regulates cell adhesion, proliferation, and apoptosis [36,37]. Additionally, they observed significant upregulation of Atf3 in IPO, the product of which is a member of the ATF/cAMP-responsive element binding protein transcription factor family and has been found to repress macrophage pro-inflammatory gene expression in IRI [38,39]. Several pathways the authors found to be differentially regulated in liver IPO were similar to those we found to be highly implicated across IRI studies. IPO significantly downregulated the MAPK and IL-17 signaling pathways, which modulate cell proliferation and apoptosis, and inflammatory cytokines and apoptosis, respectively, suggesting their role as potential protective mechanisms in IRI [18].

Caloric and Hypoxic Preconditioning
Additional conditioning methods, including caloric restriction (CR) and hypoxic preconditioning (HP), were investigated for their efficacy in attenuating renal IRI in wildtype mice [20]. Animals were subjected to either a 30% reduction in food consumption per week or three days of hypoxic exposure for 2, 4, and 8 h sequentially, prior to undergoing 40 min clamping of the left renal pedicle followed by 4, 24, or 72 h of reperfusion. CR was found to downregulate apoptotic regulator genes Bik and Dynll1 that were inversely upregulated in IRI. In non-preconditioned animals, IRI was found to upregulate similar pro-inflammatory pathways to those enriched in our KEGG analysis, including NF-κB, TNF, and IL-17. Both HP and CR were found to regulate several genes overlapping with IRI in an inverse manner, providing significant protective effects [20].

Stem Cell Therapy Approaches
Stem cell treatment was also employed to aid in reparative processes and supplement depleted cell populations in order to encourage functional recovery. Using a rodent middle cerebral artery occlusion (MCAO) model of stroke, pluripotent hematopoietic stem/progenitor cells (HSPC) were delivered 24 h post-reperfusion and were successful in reducing infarct volume and improving neurologic score [29]. HSPC treatment was found to upregulate IL-10 expression, suggesting anti-inflammatory protection by attenuating immune cell recruitment. This coincides with our findings of upregulation of pro-inflammatory IL-10 signaling in IRI. RNA sequencing also identified significantly increased metallothionein (MT-1) in transplanted cells, a cysteine-rich, free radical bindingprotein that is protective in oxidative stress [40]. Further, multiple genes involved in the pro-inflammatory DENN/ERK/MAPK pathway were found to be significantly upregulated in IRI, adding to the DEGs we found to be highly implicated in the MAPK pathway [29].
Culturing methods were compared to determine the efficacy of spheroid 3D cultures, relative to 2D, for facilitating engraftment following transplant. Multipotent human umbilical cord-derived mesenchymal stem cells (UC-MSC) were employed as a protective treatment in a rat model of hepatic IRI, using 2D and 3D cultured cells, as well as untreated controls, in order to determine an optimal treatment approach [30]. Both treatment conditions were found to reduce mRNA expression of pro-inflammatory TNF, and upregulate the expression of potentially protective IL-6 relative to the untreated controls, recapitulating the protective effects of targeting DEGs most highly implicated in IRI. Three-dimensionalcultured MSCs were found to be more effective at attenuating inflammation by acting on the TNF signaling pathway, as illustrated by their upregulation of TNF-α-induced protein 8, a potent inhibitor of apoptosis [41], and TNF-α-stimulated gene-6, a modulator of macrophage polarization [42]. Further, 3D cultured MSC were found to differentially regulate trophic factors, such as vascular endothelial growth factor, hepatocyte growth factor, and basic fibroblast growth factor, potentially affecting their subsequent engraftment and suggesting a therapeutic target to improve stem cell treatment efficacy [30].

Conclusions
In summary, IRI inflicts extensive damage by inhibiting vascular circulation, thereby disrupting metabolic and ionic regulation. This is then followed by the restoration of blood flow across oxygen-depleted areas, often overwhelming membrane flux and causing the formation of free radicals. This results in the initiation of several inflammatory kinase pathways as cells respond to ROS-induced damage and attempt to reinstate homeostasis. Several pathways are canonically dysregulated across IRI modalities, with immune signaling and cell death regulating pathways being the primary targets. Those modulating metabolic processes, ionic exchange, and developmental pathways are also highly implicated, as illustrated by functional clusters of genes displaying significant differential expression in ischemia-reperfusion injury models.
Our analysis of transcriptomic changes in IRI across organs was limited by a lack of studies providing comprehensive data on differential gene expression in IRI. Understanding the roles of these genes and pathways at large is a critical step in advancing therapeutic strategies, and the advent of RNA sequencing provides unprecedented insight into these dynamics. In this study, we analyzed differential gene expression and pathway regulation across existing studies to provide a broad overview of genes canonically implicated in IRI. Future research may apply these methods to create a more detailed, comprehensive list of transcriptomic changes in response to ischemia-reperfusion. This information may be used to guide research and optimize therapeutic approaches going forward.

Conflicts of Interest:
The authors declare no conflict of interest.